Number of days until expiry, up to 3650. Use null for never. Cannot be combined with expires_at.
expires_at
string | null
No
Exact expiry time. Must be in the future and no later than the issuing credential’s own expiry. Cannot be combined with days_to_expire.
name
string
Yes
Label for the API key (1-255 characters). API keys are not addressable by name.
policy
object
No
—
project_id
string | null
No
Unique project identifier
service_account_id
string
No
Link to an active same-organization service account. Requires an organization administrator, explicit project and nonempty policy. Omit for a personal key.
apiKeys.listList API keys (names, scopes, last used; never the secrets).Risk:read
await corsair.kernel.api.apiKeys.list({});
Input
Name
Type
Required
Description
limit
number
No
Maximum number of results to return
offset
number
No
Number of results to skip
query
string
No
Case-insensitive substring match against API key name, creator, and project. API key identifiers and masked keys match by exact value or prefix.
name
string
No
Exact-match filter on API key name using the database collation. In production, matching is case- and accent-insensitive. Names are not required to be unique, so multiple keys may match. When status=all or include_deleted=true is set, soft-deleted keys with the same name may also match.
sort_by
created_at | name | expires_at | last_used_at
No
Field to sort API keys by. Keys with no expiry or no recorded use sort last for expires_at and last_used_at in either direction.
sort_direction
asc | desc
No
Sort direction for API keys.
status
active | expired | deleted | all
No
Filter API keys by status. “active” returns keys that are neither deleted nor expired (default), “expired” returns keys that are not deleted and whose expiry has passed, “deleted” returns only soft-deleted keys, “all” returns every key.
service_account_id
string
No
Return only keys that belong to this service account.
principal_type
user | service_account
No
Return only keys that act as a user, or only keys that act as a service account.
created_by
string
No
Return only keys created by this user ID. This is the same person the key’s created_by field describes.
project_id
string
No
Return only keys scoped to this project ID. The value “org” returns keys that are not scoped to a project.
include_deleted
boolean
No
Deprecated: use status=all instead. When true, include deleted (soft-deleted) API keys in the results for audit purposes.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
apiKeys.rotateReplace an API key’s secret. The new secret is shown once. The old one keeps working for expire_in_days (7 by default); pass expire_in_days: 0 to end it now.Risk:destructive · Irreversible
auditLogs.exportChunkDownload one chunk of the audit log as JSONL (gzipped by default). Returns a file plus has_more and next_cursor for the next chunk.Risk:read
Lower bound (inclusive) for the audit record timestamp.
end
string
Yes
Upper bound (exclusive) for the audit record timestamp.
auth_strategy
string
No
Filter by authentication strategy.
service
string
No
Filter by service name.
method
string
No
Filter by HTTP method.
exclude_method
string[]
No
Filter out results by HTTP method.
search
string
No
Free-text search over path, user ID, email, client IP, and status.
search_user_id
string[]
No
Additional user IDs to OR into free-text search.
cursor
string
No
Opaque cursor from X-Next-Cursor for the next chunk of older records.
limit
number
No
Maximum number of records to return in this chunk.
format
jsonl | jsonl.gz
No
Encoding for the returned chunk.
Output
Name
Type
Required
Description
contentBase64
string
No
The file, base64-encoded. Missing when the file came back as text or contentOmitted is set.
text
string
No
The file as text, in place of contentBase64, when it is a text file (CSV, JSON, Markdown, XML and the like) in UTF-8.
contentType
string
Yes
Media type of the file, for example application/pdf.
byteLength
number
Yes
Size of the file in bytes.
filename
string
No
The file name, when known.
contentOmitted
not_requested | over_limit
No
Set when the content was left out: the caller did not ask for it, or the file was over the limit.
fileId
string
No
Where the host saved the file. Over the limit it comes in place of the content; a host that keeps every file sets it beside the content too. On Corsair it is a Files id (fil_…): pass it to another app wherever that app takes a file URL, or read a text file or PDF with files.read_text.
has_more
boolean
Yes
True when another chunk follows; pass next_cursor as cursor to read it.
next_cursor
string | null
Yes
Pass as cursor to read the next chunk.
exportDestinations.create
auditLogs.exportDestinations.createAdd a destination (such as an S3 bucket) that receives continuous audit log exports.Risk:write
The Kernel role that assumes role_arn in your account to deliver logs. Allow this role as the principal in your role’s trust policy, and require external_id as the sts:ExternalId condition. Recreating a destination issues a new external_id, which the trust policy has to be updated to match.
kms_key_id
string | null
No
—
last_error
string | null
No
Sanitized description of the most recent delivery failure.
last_error_at
string | null
No
—
last_exported_cursor
string | null
No
Opaque, versioned checkpoint for forward-only continuous export. This value is not compatible with audit-log list page tokens. Delivery starts at the moment the destination is activated, so events recorded before that are not delivered. Pausing stops delivery and resuming starts again from the time of the resume: events recorded while a destination was paused are never exported, and pausing is not a way to defer delivery.
last_success_at
string | null
No
—
next_attempt_at
string | null
No
—
prefix
string | null
No
—
region
string | null
No
—
role_arn
string | null
No
—
status
string | null
No
Pausing prevents new delivery attempts. An S3 upload already in progress may complete after the pause response; its rows can appear again after the destination is resumed.
type
string | null
No
—
updated_at
string | null
No
—
exportDestinations.delete
auditLogs.exportDestinations.deleteDelete an audit log export destination. Cannot be undone.Risk:destructive · Irreversible
The Kernel role that assumes role_arn in your account to deliver logs. Allow this role as the principal in your role’s trust policy, and require external_id as the sts:ExternalId condition. Recreating a destination issues a new external_id, which the trust policy has to be updated to match.
kms_key_id
string | null
No
—
last_error
string | null
No
Sanitized description of the most recent delivery failure.
last_error_at
string | null
No
—
last_exported_cursor
string | null
No
Opaque, versioned checkpoint for forward-only continuous export. This value is not compatible with audit-log list page tokens. Delivery starts at the moment the destination is activated, so events recorded before that are not delivered. Pausing stops delivery and resuming starts again from the time of the resume: events recorded while a destination was paused are never exported, and pausing is not a way to defer delivery.
last_success_at
string | null
No
—
next_attempt_at
string | null
No
—
prefix
string | null
No
—
region
string | null
No
—
role_arn
string | null
No
—
status
string | null
No
Pausing prevents new delivery attempts. An S3 upload already in progress may complete after the pause response; its rows can appear again after the destination is resumed.
KMS key ID, alias, or ARN. Set to an empty string to remove the configured KMS key; omit or send null to leave unchanged.
prefix
string
No
—
region
string
No
—
role_arn
string
No
—
status
active | paused
No
—
Output
Name
Type
Required
Description
bucket
string | null
No
—
consecutive_failures
number | null
No
—
created_at
string | null
No
—
external_id
string | null
No
—
format
string | null
No
—
id
string | null
No
—
kernel_role_arn
string | null
No
The Kernel role that assumes role_arn in your account to deliver logs. Allow this role as the principal in your role’s trust policy, and require external_id as the sts:ExternalId condition. Recreating a destination issues a new external_id, which the trust policy has to be updated to match.
kms_key_id
string | null
No
—
last_error
string | null
No
Sanitized description of the most recent delivery failure.
last_error_at
string | null
No
—
last_exported_cursor
string | null
No
Opaque, versioned checkpoint for forward-only continuous export. This value is not compatible with audit-log list page tokens. Delivery starts at the moment the destination is activated, so events recorded before that are not delivered. Pausing stops delivery and resuming starts again from the time of the resume: events recorded while a destination was paused are never exported, and pausing is not a way to defer delivery.
last_success_at
string | null
No
—
next_attempt_at
string | null
No
—
prefix
string | null
No
—
region
string | null
No
—
role_arn
string | null
No
—
status
string | null
No
Pausing prevents new delivery attempts. An S3 upload already in progress may complete after the pause response; its rows can appear again after the destination is resumed.
type
string | null
No
—
updated_at
string | null
No
—
list
auditLogs.listSearch the org’s audit log (who called what, when), newest first. Pages with page_token; check has_more.Risk:read
auth.connections.createSet up managed login for a site: one connection per profile and domain (the profile is created if missing). Then start a login with auth.connections.login.Risk:write
Additional hostname roots valid for this auth flow, besides the primary domain. Each value allows credential entry on that exact hostname and its subdomains. Leading www. and *. labels are normalized away. When omitted or empty, credential entry is unrestricted. The following SSO/OAuth provider domains are automatically allowed by default and do not need to be specified: - Google: accounts.google.com - Microsoft/Azure AD: login.microsoftonline.com, login.live.com - Okta: *.okta.com, *.oktapreview.com - Auth0: *.auth0.com, *.us.auth0.com, *.eu.auth0.com, *.au.auth0.com - Apple: appleid.apple.com - GitHub: github.com - Facebook/Meta: www.facebook.com - LinkedIn: www.linkedin.com - Amazon Cognito: *.amazoncognito.com - OneLogin: *.onelogin.com - Ping Identity: *.pingone.com, *.pingidentity.com
auto_reauth
boolean
No
Whether to permit automatic re-authentication when a scheduled health check detects an expired session. This is an opt-in flag only — it does not check whether re-auth is actually feasible. Even when true, re-auth only runs when the system has what it needs to perform it (for example, saved credentials for the required login fields), and only after a scheduled health check detects an expired session — so this flag has no effect when health_checks is false. When false, expired sessions are marked as NEEDS_AUTH instead of attempting re-auth. Defaults to true.
browser
object
No
—
browser_telemetry
object | null
No
—
credential
object
No
—
domain
string
Yes
Domain for authentication
health_check_interval
number
No
Interval in seconds between automatic health checks. When set, the system periodically verifies the authentication status and triggers re-authentication if needed. Maximum is 86400 (24 hours). Default is 3600 (1 hour) or your plan minimum, whichever is larger. The minimum depends on your plan: Enterprise: 300 (5 minutes), Startup: 1200 (20 minutes), Hobbyist: 3600 (1 hour), Free: 21600 (6 hours).
health_checks
boolean
No
Whether to enable periodic health checks. When false, the system will not automatically verify authentication status, and auto_reauth has no effect on the automatic flow (since re-auth is only triggered by a failed scheduled health check). Defaults to true.
hitl
boolean
No
Whether a login blocked by a CAPTCHA the browser cannot solve pauses for a person instead of failing. The flow moves to AWAITING_HUMAN_INTERVENTION with a human_intervention choice; the person completes the challenge in live_view_url, then submits that choice to resume. Reauthentication never pauses. Defaults to false.
login_url
string
No
Optional login page URL to skip discovery
profile_name
string
Yes
Name of the profile to manage authentication for. If the profile does not exist, it is created automatically.
proxy
object
No
—
record_session
boolean
No
Whether to record browser sessions for this connection by default. Useful for debugging. Can be overridden per-login. Defaults to false.
save_credentials
boolean
No
Whether to save credentials after every successful login. Defaults to true. One-time codes (TOTP, SMS, etc.) are not saved.
Additional hostname roots valid for this auth flow, besides the primary domain. Each value allows credential entry on that exact hostname and its subdomains. Leading www. and *. labels are normalized away. When omitted or empty, credential entry is unrestricted. The following SSO/OAuth provider domains are automatically allowed by default and do not need to be specified: - Google: accounts.google.com - Microsoft/Azure AD: login.microsoftonline.com, login.live.com - Okta: *.okta.com, *.oktapreview.com - Auth0: *.auth0.com, *.us.auth0.com, *.eu.auth0.com, *.au.auth0.com - Apple: appleid.apple.com - GitHub: github.com - Facebook/Meta: www.facebook.com - LinkedIn: www.linkedin.com - Amazon Cognito: *.amazoncognito.com - OneLogin: *.onelogin.com - Ping Identity: *.pingone.com, *.pingidentity.com
auto_reauth
boolean | null
No
Whether automatic re-authentication is permitted for this connection. This is an opt-in flag only — it does not check whether re-auth is actually feasible. Even when true, re-auth only runs when the system has what it needs to perform it (for example, saved credentials for the required login fields), and only after a scheduled health check detects an expired session — so this flag has no effect when health_checks is false. When false, expired sessions detected by a health check are marked as NEEDS_AUTH instead of attempting re-auth.
browser
object | null
No
—
browser_session_id
string | null | null
No
ID of the underlying browser session driving the current flow (present when flow in progress). Use this to inspect or terminate the browser session via the /browsers API.
browser_telemetry
object | null
No
—
can_reauth
boolean | null
No
Whether this connection’s stored requirements are eligible for unattended re-authentication. A true value can represent either fully satisfiable requirements or a best-effort attempt. It does not account for whether automatic re-authentication is enabled or parent workflow state such as an active flow or circuit-breaker cooldown, so it does not guarantee an attempt on the next health check. See can_reauth_reason for the specific outcome.
can_reauth_reason
string | null
No
Machine-readable reason for the current value of can_reauth. Affirmative values (requirements are eligible for unattended re-authentication): - external_credential — an external credential provider is attached - cua_has_credential — CUA flow with a stored credential - has_credential — Kernel credential is attached (optimistic; plan viability not checked) - viable_plans_found — at least one stored login plan can be replayed - no_requirements_recorded — no recorded credential requirements to fail against - totp_reauth_allowed — TOTP is the only recorded requirement and a stored secret can generate the code - optimistic_totp_attempt — TOTP was previously required but no reusable secret is stored; the connection remains eligible for a circuit-breaker-bounded attempt because the site may not challenge returning sessions - requirements_satisfiable — recorded requirements contain no recognized blocker Negative values (a human must complete the login flow): - no_prior_successful_login — connection has never completed a successful login - no_credential — no Kernel or external credential attached - no_viable_plans — credential attached but no replayable login plan exists yet - viable_plans_require_external_action — stored plans need an external step (email link, push, etc.) - requires_external_action — recorded requirements include an external step - requires_totp_without_secret — flow needs a TOTP code but no TOTP secret is stored - requires_sms_code — flow needs an SMS code that cannot be received automatically - requires_email_code — flow needs an email code that cannot be received automatically - requires_customer_input — flow needs another field or choice that is unavailable during unattended re-authentication
choices
object[] | null
No
Canonical choices awaiting selection. Prefer this over pending_sso_buttons, mfa_options, and sign_in_options when present.
credential
object | null
No
—
discovered_fields
object[] | null
No
Fields awaiting input (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
domain
string | null
No
Target domain for authentication
error_code
string | null | null
No
Machine-readable error code (present when flow_status=failed)
error_message
string | null | null
No
Error message (present when flow_status=failed)
external_action_message
string | null | null
No
Instructions for external action (present when flow_step=awaiting_external_action)
fields
object[] | null
No
Canonical fields awaiting input. Prefer this over discovered_fields when present.
flow_expires_at
string | null | null
No
When the current flow expires (null when no flow in progress). A flow past this timestamp is no longer valid and its flow_status will be EXPIRED. Clients may start a new login to supersede a stale IN_PROGRESS flow past this timestamp.
flow_status
string | null | null
No
Current flow status (null when no flow in progress)
flow_step
string | null | null
No
Current step in the flow (null when no flow in progress)
flow_type
string | null | null
No
Type of the current flow (null when no flow in progress)
health_check_interval
number | null | null
No
Interval in seconds between automatic health checks. When set, the system periodically verifies the authentication status and triggers re-authentication if needed. Maximum is 86400 (24 hours). Default is 3600 (1 hour) or your plan minimum, whichever is larger. The minimum depends on your plan: Enterprise: 300 (5 minutes), Startup: 1200 (20 minutes), Hobbyist: 3600 (1 hour), Free: 21600 (6 hours).
health_check_unavailable_reason
string | null
No
Why health checks cannot verify this connection. Present when health checks are enabled but no auth check URL is available; a recent last_auth_check_at is not evidence of a valid session.
health_checks
boolean | null
No
Whether periodic health checks are enabled for this connection. When false, the system will not automatically verify authentication status, and auto_reauth has no effect on the automatic flow (since re-auth is only triggered by a failed scheduled health check). Manually triggering a health check via the API still works regardless of this setting.
hitl
boolean | null
No
Whether a login blocked by a CAPTCHA the browser cannot solve pauses at AWAITING_HUMAN_INTERVENTION so a person can complete it in the live view.
hosted_url
string | null | null
No
URL to redirect user to for hosted login (present when flow in progress)
id
string | null
No
Unique identifier for the auth connection
interaction_id
string | null | null
No
Opaque identifier for the current canonical interaction. Required when submitting fields or choices and changes for each new actionable pause.
last_auth_at
string | null
No
Deprecated alias for last_auth_check_at. Despite the name, this is the last health-check timestamp, not the last successful authentication. Use last_auth_check_at instead.
last_auth_check_at
string | null
No
When the most recent auth health check ran for this connection, regardless of outcome. Updated on every health check and does not by itself indicate that the profile is currently authenticated - use status for that. May be newer than flow_expires_at when a flow is still in progress because health checks continue to run in parallel.
live_view_url
string | null | null
No
Browser live view URL for debugging (present when flow in progress)
login_url
string | null
No
Optional login page URL to skip discovery
mfa_options
object[] | null
No
MFA method options (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
pending_sso_buttons
object[] | null
No
SSO buttons available (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
post_login_url
string | null
No
URL where the browser landed after successful login
profile_name
string | null
No
Name of the profile associated with this auth connection
proxy_id
string | null
No
Deprecated. Read browser.proxy instead. Retained during migration for existing clients.
record_session
boolean | null
No
Whether to record browser session replays for this connection by default. Useful for debugging login flows. Can be overridden per-login.
save_credentials
boolean | null
No
Whether credentials are saved after every successful login. One-time codes (TOTP, SMS, etc.) are not saved.
sign_in_options
object[] | null
No
Non-MFA choices presented during the auth flow, such as account selection or org pickers (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions).
sso_provider
string | null | null
No
SSO provider being used (e.g., google, github, microsoft)
status
string | null
No
Last known authentication status of the managed profile. An inconclusive health check preserves this status and does not verify the current session.
website_error
string | null | null
No
Visible error message from the website (e.g., ‘Incorrect password’). Present when the website displays an error during login.
auth.connections.deleteDelete a managed auth connection and stop its health checks and logins. Cannot be undone; the profile stays.Risk:destructive · Irreversible
JWT token with invocation_id claim (30 minute TTL)
connections.get
auth.connections.getGet a managed auth connection and its current login flow: status, step, the fields it is waiting for, and hosted_url or live_view_url for a person to finish it. Poll this while a login runs.Risk:read
Additional hostname roots valid for this auth flow, besides the primary domain. Each value allows credential entry on that exact hostname and its subdomains. Leading www. and *. labels are normalized away. When omitted or empty, credential entry is unrestricted. The following SSO/OAuth provider domains are automatically allowed by default and do not need to be specified: - Google: accounts.google.com - Microsoft/Azure AD: login.microsoftonline.com, login.live.com - Okta: *.okta.com, *.oktapreview.com - Auth0: *.auth0.com, *.us.auth0.com, *.eu.auth0.com, *.au.auth0.com - Apple: appleid.apple.com - GitHub: github.com - Facebook/Meta: www.facebook.com - LinkedIn: www.linkedin.com - Amazon Cognito: *.amazoncognito.com - OneLogin: *.onelogin.com - Ping Identity: *.pingone.com, *.pingidentity.com
auto_reauth
boolean | null
No
Whether automatic re-authentication is permitted for this connection. This is an opt-in flag only — it does not check whether re-auth is actually feasible. Even when true, re-auth only runs when the system has what it needs to perform it (for example, saved credentials for the required login fields), and only after a scheduled health check detects an expired session — so this flag has no effect when health_checks is false. When false, expired sessions detected by a health check are marked as NEEDS_AUTH instead of attempting re-auth.
browser
object | null
No
—
browser_session_id
string | null | null
No
ID of the underlying browser session driving the current flow (present when flow in progress). Use this to inspect or terminate the browser session via the /browsers API.
browser_telemetry
object | null
No
—
can_reauth
boolean | null
No
Whether this connection’s stored requirements are eligible for unattended re-authentication. A true value can represent either fully satisfiable requirements or a best-effort attempt. It does not account for whether automatic re-authentication is enabled or parent workflow state such as an active flow or circuit-breaker cooldown, so it does not guarantee an attempt on the next health check. See can_reauth_reason for the specific outcome.
can_reauth_reason
string | null
No
Machine-readable reason for the current value of can_reauth. Affirmative values (requirements are eligible for unattended re-authentication): - external_credential — an external credential provider is attached - cua_has_credential — CUA flow with a stored credential - has_credential — Kernel credential is attached (optimistic; plan viability not checked) - viable_plans_found — at least one stored login plan can be replayed - no_requirements_recorded — no recorded credential requirements to fail against - totp_reauth_allowed — TOTP is the only recorded requirement and a stored secret can generate the code - optimistic_totp_attempt — TOTP was previously required but no reusable secret is stored; the connection remains eligible for a circuit-breaker-bounded attempt because the site may not challenge returning sessions - requirements_satisfiable — recorded requirements contain no recognized blocker Negative values (a human must complete the login flow): - no_prior_successful_login — connection has never completed a successful login - no_credential — no Kernel or external credential attached - no_viable_plans — credential attached but no replayable login plan exists yet - viable_plans_require_external_action — stored plans need an external step (email link, push, etc.) - requires_external_action — recorded requirements include an external step - requires_totp_without_secret — flow needs a TOTP code but no TOTP secret is stored - requires_sms_code — flow needs an SMS code that cannot be received automatically - requires_email_code — flow needs an email code that cannot be received automatically - requires_customer_input — flow needs another field or choice that is unavailable during unattended re-authentication
choices
object[] | null
No
Canonical choices awaiting selection. Prefer this over pending_sso_buttons, mfa_options, and sign_in_options when present.
credential
object | null
No
—
discovered_fields
object[] | null
No
Fields awaiting input (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
domain
string | null
No
Target domain for authentication
error_code
string | null | null
No
Machine-readable error code (present when flow_status=failed)
error_message
string | null | null
No
Error message (present when flow_status=failed)
external_action_message
string | null | null
No
Instructions for external action (present when flow_step=awaiting_external_action)
fields
object[] | null
No
Canonical fields awaiting input. Prefer this over discovered_fields when present.
flow_expires_at
string | null | null
No
When the current flow expires (null when no flow in progress). A flow past this timestamp is no longer valid and its flow_status will be EXPIRED. Clients may start a new login to supersede a stale IN_PROGRESS flow past this timestamp.
flow_status
string | null | null
No
Current flow status (null when no flow in progress)
flow_step
string | null | null
No
Current step in the flow (null when no flow in progress)
flow_type
string | null | null
No
Type of the current flow (null when no flow in progress)
health_check_interval
number | null | null
No
Interval in seconds between automatic health checks. When set, the system periodically verifies the authentication status and triggers re-authentication if needed. Maximum is 86400 (24 hours). Default is 3600 (1 hour) or your plan minimum, whichever is larger. The minimum depends on your plan: Enterprise: 300 (5 minutes), Startup: 1200 (20 minutes), Hobbyist: 3600 (1 hour), Free: 21600 (6 hours).
health_check_unavailable_reason
string | null
No
Why health checks cannot verify this connection. Present when health checks are enabled but no auth check URL is available; a recent last_auth_check_at is not evidence of a valid session.
health_checks
boolean | null
No
Whether periodic health checks are enabled for this connection. When false, the system will not automatically verify authentication status, and auto_reauth has no effect on the automatic flow (since re-auth is only triggered by a failed scheduled health check). Manually triggering a health check via the API still works regardless of this setting.
hitl
boolean | null
No
Whether a login blocked by a CAPTCHA the browser cannot solve pauses at AWAITING_HUMAN_INTERVENTION so a person can complete it in the live view.
hosted_url
string | null | null
No
URL to redirect user to for hosted login (present when flow in progress)
id
string | null
No
Unique identifier for the auth connection
interaction_id
string | null | null
No
Opaque identifier for the current canonical interaction. Required when submitting fields or choices and changes for each new actionable pause.
last_auth_at
string | null
No
Deprecated alias for last_auth_check_at. Despite the name, this is the last health-check timestamp, not the last successful authentication. Use last_auth_check_at instead.
last_auth_check_at
string | null
No
When the most recent auth health check ran for this connection, regardless of outcome. Updated on every health check and does not by itself indicate that the profile is currently authenticated - use status for that. May be newer than flow_expires_at when a flow is still in progress because health checks continue to run in parallel.
live_view_url
string | null | null
No
Browser live view URL for debugging (present when flow in progress)
login_url
string | null
No
Optional login page URL to skip discovery
mfa_options
object[] | null
No
MFA method options (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
pending_sso_buttons
object[] | null
No
SSO buttons available (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
post_login_url
string | null
No
URL where the browser landed after successful login
profile_name
string | null
No
Name of the profile associated with this auth connection
proxy_id
string | null
No
Deprecated. Read browser.proxy instead. Retained during migration for existing clients.
record_session
boolean | null
No
Whether to record browser session replays for this connection by default. Useful for debugging login flows. Can be overridden per-login.
save_credentials
boolean | null
No
Whether credentials are saved after every successful login. One-time codes (TOTP, SMS, etc.) are not saved.
sign_in_options
object[] | null
No
Non-MFA choices presented during the auth flow, such as account selection or org pickers (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions).
sso_provider
string | null | null
No
SSO provider being used (e.g., google, github, microsoft)
status
string | null
No
Last known authentication status of the managed profile. An inconclusive health check preserves this status and does not verify the current session.
website_error
string | null | null
No
Visible error message from the website (e.g., ‘Incorrect password’). Present when the website displays an error during login.
auth.connections.loginStart a login for a managed auth connection. Returns hosted_url (and live_view_url) for a person to complete it, or logs in by itself when credentials are stored. Then poll auth.connections.get, and send asked-for fields with auth.connections.submit.Risk:write
Override the connection’s default for recording this login’s browser session. When omitted, the connection’s record_session default is used.
skill_mode
enabled | disabled
No
Controls whether this login reads and writes learned domain skills. Automatic reauths inherit the selected mode until a later accepted login sets enabled or omits this field. Defaults to enabled when omitted.
auth.connections.submitSend the values a running login asks for (a username, password, one-time code, or a choice) to a managed auth connection. Poll auth.connections.get for the result.Risk:write
Additional hostname roots valid for this auth flow. Each value allows credential entry on that exact hostname and its subdomains; leading www. and *. labels are normalized away. An empty list leaves credential entry unrestricted. Replaces the existing list.
auto_reauth
boolean
No
Whether automatic re-authentication is permitted for this connection. This is an opt-in flag only — it does not check whether re-auth is actually feasible. Even when true, re-auth only runs when the system has what it needs to perform it (for example, saved credentials for the required login fields), and only after a scheduled health check detects an expired session — so this flag has no effect when health_checks is false. When false, expired sessions detected by a health check are marked as NEEDS_AUTH instead of attempting re-auth.
browser
object
No
—
browser_telemetry
object | null
No
—
credential
object
No
—
health_check_interval
number
No
Interval in seconds between automatic health checks
health_checks
boolean
No
Whether periodic health checks are enabled. When set to false, the system will not automatically verify authentication status, and auto_reauth has no effect on the automatic flow (since re-auth is only triggered by a failed scheduled health check).
hitl
boolean
No
Whether a login blocked by a CAPTCHA the browser cannot solve pauses for a person instead of failing
login_url
string
No
Login page URL. Set to empty string to clear.
proxy
object
No
—
record_session
boolean
No
Whether to record browser sessions for this connection by default
save_credentials
boolean
No
Whether to save credentials after every successful login
Additional hostname roots valid for this auth flow, besides the primary domain. Each value allows credential entry on that exact hostname and its subdomains. Leading www. and *. labels are normalized away. When omitted or empty, credential entry is unrestricted. The following SSO/OAuth provider domains are automatically allowed by default and do not need to be specified: - Google: accounts.google.com - Microsoft/Azure AD: login.microsoftonline.com, login.live.com - Okta: *.okta.com, *.oktapreview.com - Auth0: *.auth0.com, *.us.auth0.com, *.eu.auth0.com, *.au.auth0.com - Apple: appleid.apple.com - GitHub: github.com - Facebook/Meta: www.facebook.com - LinkedIn: www.linkedin.com - Amazon Cognito: *.amazoncognito.com - OneLogin: *.onelogin.com - Ping Identity: *.pingone.com, *.pingidentity.com
auto_reauth
boolean | null
No
Whether automatic re-authentication is permitted for this connection. This is an opt-in flag only — it does not check whether re-auth is actually feasible. Even when true, re-auth only runs when the system has what it needs to perform it (for example, saved credentials for the required login fields), and only after a scheduled health check detects an expired session — so this flag has no effect when health_checks is false. When false, expired sessions detected by a health check are marked as NEEDS_AUTH instead of attempting re-auth.
browser
object | null
No
—
browser_session_id
string | null | null
No
ID of the underlying browser session driving the current flow (present when flow in progress). Use this to inspect or terminate the browser session via the /browsers API.
browser_telemetry
object | null
No
—
can_reauth
boolean | null
No
Whether this connection’s stored requirements are eligible for unattended re-authentication. A true value can represent either fully satisfiable requirements or a best-effort attempt. It does not account for whether automatic re-authentication is enabled or parent workflow state such as an active flow or circuit-breaker cooldown, so it does not guarantee an attempt on the next health check. See can_reauth_reason for the specific outcome.
can_reauth_reason
string | null
No
Machine-readable reason for the current value of can_reauth. Affirmative values (requirements are eligible for unattended re-authentication): - external_credential — an external credential provider is attached - cua_has_credential — CUA flow with a stored credential - has_credential — Kernel credential is attached (optimistic; plan viability not checked) - viable_plans_found — at least one stored login plan can be replayed - no_requirements_recorded — no recorded credential requirements to fail against - totp_reauth_allowed — TOTP is the only recorded requirement and a stored secret can generate the code - optimistic_totp_attempt — TOTP was previously required but no reusable secret is stored; the connection remains eligible for a circuit-breaker-bounded attempt because the site may not challenge returning sessions - requirements_satisfiable — recorded requirements contain no recognized blocker Negative values (a human must complete the login flow): - no_prior_successful_login — connection has never completed a successful login - no_credential — no Kernel or external credential attached - no_viable_plans — credential attached but no replayable login plan exists yet - viable_plans_require_external_action — stored plans need an external step (email link, push, etc.) - requires_external_action — recorded requirements include an external step - requires_totp_without_secret — flow needs a TOTP code but no TOTP secret is stored - requires_sms_code — flow needs an SMS code that cannot be received automatically - requires_email_code — flow needs an email code that cannot be received automatically - requires_customer_input — flow needs another field or choice that is unavailable during unattended re-authentication
choices
object[] | null
No
Canonical choices awaiting selection. Prefer this over pending_sso_buttons, mfa_options, and sign_in_options when present.
credential
object | null
No
—
discovered_fields
object[] | null
No
Fields awaiting input (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
domain
string | null
No
Target domain for authentication
error_code
string | null | null
No
Machine-readable error code (present when flow_status=failed)
error_message
string | null | null
No
Error message (present when flow_status=failed)
external_action_message
string | null | null
No
Instructions for external action (present when flow_step=awaiting_external_action)
fields
object[] | null
No
Canonical fields awaiting input. Prefer this over discovered_fields when present.
flow_expires_at
string | null | null
No
When the current flow expires (null when no flow in progress). A flow past this timestamp is no longer valid and its flow_status will be EXPIRED. Clients may start a new login to supersede a stale IN_PROGRESS flow past this timestamp.
flow_status
string | null | null
No
Current flow status (null when no flow in progress)
flow_step
string | null | null
No
Current step in the flow (null when no flow in progress)
flow_type
string | null | null
No
Type of the current flow (null when no flow in progress)
health_check_interval
number | null | null
No
Interval in seconds between automatic health checks. When set, the system periodically verifies the authentication status and triggers re-authentication if needed. Maximum is 86400 (24 hours). Default is 3600 (1 hour) or your plan minimum, whichever is larger. The minimum depends on your plan: Enterprise: 300 (5 minutes), Startup: 1200 (20 minutes), Hobbyist: 3600 (1 hour), Free: 21600 (6 hours).
health_check_unavailable_reason
string | null
No
Why health checks cannot verify this connection. Present when health checks are enabled but no auth check URL is available; a recent last_auth_check_at is not evidence of a valid session.
health_checks
boolean | null
No
Whether periodic health checks are enabled for this connection. When false, the system will not automatically verify authentication status, and auto_reauth has no effect on the automatic flow (since re-auth is only triggered by a failed scheduled health check). Manually triggering a health check via the API still works regardless of this setting.
hitl
boolean | null
No
Whether a login blocked by a CAPTCHA the browser cannot solve pauses at AWAITING_HUMAN_INTERVENTION so a person can complete it in the live view.
hosted_url
string | null | null
No
URL to redirect user to for hosted login (present when flow in progress)
id
string | null
No
Unique identifier for the auth connection
interaction_id
string | null | null
No
Opaque identifier for the current canonical interaction. Required when submitting fields or choices and changes for each new actionable pause.
last_auth_at
string | null
No
Deprecated alias for last_auth_check_at. Despite the name, this is the last health-check timestamp, not the last successful authentication. Use last_auth_check_at instead.
last_auth_check_at
string | null
No
When the most recent auth health check ran for this connection, regardless of outcome. Updated on every health check and does not by itself indicate that the profile is currently authenticated - use status for that. May be newer than flow_expires_at when a flow is still in progress because health checks continue to run in parallel.
live_view_url
string | null | null
No
Browser live view URL for debugging (present when flow in progress)
login_url
string | null
No
Optional login page URL to skip discovery
mfa_options
object[] | null
No
MFA method options (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
pending_sso_buttons
object[] | null
No
SSO buttons available (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions)
post_login_url
string | null
No
URL where the browser landed after successful login
profile_name
string | null
No
Name of the profile associated with this auth connection
proxy_id
string | null
No
Deprecated. Read browser.proxy instead. Retained during migration for existing clients.
record_session
boolean | null
No
Whether to record browser session replays for this connection by default. Useful for debugging login flows. Can be overridden per-login.
save_credentials
boolean | null
No
Whether credentials are saved after every successful login. One-time codes (TOTP, SMS, etc.) are not saved.
sign_in_options
object[] | null
No
Non-MFA choices presented during the auth flow, such as account selection or org pickers (present when flow_step=awaiting_input; may also be present with awaiting_external_action as fallback actions).
sso_provider
string | null | null
No
SSO provider being used (e.g., google, github, microsoft)
status
string | null
No
Last known authentication status of the managed profile. An inconclusive health check preserves this status and does not verify the current session.
website_error
string | null | null
No
Visible error message from the website (e.g., ‘Incorrect password’). Present when the website displays an error during login.
browserPools.acquireTake a ready browser from a pool (waits up to acquire_timeout_seconds). Returns null when none freed up in time; call again. Give it back with browserPools.release.Risk:write
Maximum number of seconds to wait for a browser to be available. Defaults to the calculated time it would take to fill the pool at the currently configured fill rate.
name
string
No
Optional human-readable name for the acquired browser session, used to find it later in the dashboard. Must be unique among active sessions within the pool’s project. Applies to this lease only and is cleared when the browser is released back to the pool.
profile
object
No
Optional profile to load before returning the acquired browser. The pool itself must not already have a profile configured. A browser loaded with an acquire-time profile is destroyed and replaced on release instead of being reused, so profile state cannot leak into another lease. Set save_changes to persist changes when that browser is destroyed.
start_replay
boolean
No
Start replay recording immediately for the acquired browser. Omit to use the pool’s configured default.
start_url
string
No
Optional URL to navigate the acquired browser to. Overrides the pool’s start_url for this acquire only. Best-effort: failures to navigate do not fail the acquire.
Remote URL for live viewing the browser session. Only available for non-headless browsers.
captcha_solver
object | null
No
—
cdp_ws_url
string | null
No
Websocket URL for Chrome DevTools Protocol connections to the browser session
chrome_policy
object | null
No
Custom Chrome enterprise policy overrides that were applied to this browser session, if any. Echoed back for verification. Keys are Chrome enterprise policy names.
cost
object | null
No
—
created_at
string | null
No
When the browser session was created.
deleted_at
string | null
No
When the browser session was soft-deleted. Only present for deleted sessions.
fingerprint
object | null
No
—
gpu
boolean | null
No
Whether GPU acceleration is enabled for the browser session (only supported for headful sessions).
headless
boolean | null
No
Whether the browser session is running in headless mode.
kiosk_mode
boolean | null
No
Whether the browser session is running in kiosk mode.
leased_at
string | null
No
When the pooled browser session was last leased. Only present for leased pooled sessions.
location
object | null
No
—
memory
string | null
No
—
name
string | null
No
Human-readable name of the browser session, if one was set at creation.
network
object | null
No
—
pool
object | null
No
—
profile
object | null
No
—
profile_save_changes
boolean | null
No
Whether changes made during this browser session are saved back to its profile when the session ends. Omitted when no profile is attached.
proxy
object | null
No
—
proxy_id
string | null
No
ID of the proxy associated with this browser session, if any. Deprecated in favor of proxy.
region
string | null
No
—
replay
object | null
No
—
session_id
string | null
No
Unique identifier for the browser session
start_url
string | null
No
URL the session was most recently asked to navigate to, if any. Recorded for debugging. Navigation is fire-and-forget — the URL is dispatched to the browser without waiting for it to load, and any errors (DNS failure, bad status, timeout) are silently dropped. Captures what was requested, not what the browser actually loaded.
stealth
boolean | null
No
Whether the browser session has CAPTCHA solving enabled, in which case proxy.mode=default uses Kernel’s default stealth proxy. A browser created with only captcha_solver reports true with proxy.mode=direct. Deprecated in favor of captcha_solver and proxy.
tags
Record<string, string> | null
No
—
telemetry
object | null
No
—
timeout_seconds
number | null
No
The number of seconds of inactivity before the browser session is terminated.
usage
object | null
No
—
usage_status
string | null
No
—
vaults
object[] | null
No
Vaults linked when the browser session was created.
video_memory
string | null
No
—
viewport
object | null
No
—
webdriver_ws_url
string | null
No
Websocket URL for WebDriver BiDi connections to the browser session
{ height?: number | null, refresh_rate?: number | null, width?: number | null} | null
create
browserPools.createCreate a pool of pre-started browsers with one configuration, for instant browserPools.acquire. Idle pool browsers are not billed but count toward the concurrency limit.Risk:write
Custom Chrome enterprise policy overrides applied to all browsers in this pool. Keys are Chrome enterprise policy names; values must match their expected types. Blocked: kernel-managed policies (extensions, proxy, CDP/automation). See https://chromeenterprise.google/policies/ The serialized JSON payload is capped at 5 MiB.
extensions
object[]
No
List of browser extensions to load into the session. Provide each by id or name.
fill_rate_per_minute
number
No
Percentage of the pool to fill per minute. Defaults to 25. The cap is 25 for most organizations but can be raised per-organization, so only the lower bound is enforced here.
headless
boolean
No
If true, launches the browser using a headless image. Defaults to false.
kiosk_mode
boolean
No
If true, launches the browser in kiosk mode to hide address bar and tabs in live view. Defaults to false.
memory
1GiB | 2GiB | 4GiB | 8GiB | 16GiB
No
—
name
string
No
Optional name for the browser pool. Must be unique within the project.
network
object
No
—
profile
object
No
Profile selection for browsers in the pool.
proxy_id
string
No
Optional proxy to associate to the browser session. Must reference a proxy in the same project as the browser session.
refresh_on_profile_update
boolean
No
When true, flush idle browsers when the profile the pool uses is updated, so pool browsers pick up the latest profile data. When a profile is provided during creation, this defaults to true. Requires a profile to be set on the pool.
region
us-east | us-west | eu-west | ap-southeast
No
—
size
number
Yes
Number of browsers to maintain in the pool. The maximum size is determined by your organization’s pooled sessions limit (the sum of all pool sizes cannot exceed your limit).
start_replay
boolean
No
Start replay recording automatically when a browser is acquired from this pool.
start_url
string
No
Optional URL to navigate to when a new browser is warmed into the pool. Best-effort: failures to navigate do not fail pool fill. Only applied to newly-warmed browsers; browsers reused via release/acquire keep whatever URL the previous lease left them on. Accepts any URL Chromium can resolve, including chrome:// pages.
stealth
boolean
No
If true, enables the CAPTCHA solver and, when proxy_id is omitted, uses Kernel’s default stealth proxy. Defaults to false. Cannot be combined with captcha_solver.enabled=false. Deprecated in favor of captcha_solver and proxy_id.
telemetry
object | null
No
—
timeout_seconds
number
No
Default idle timeout in seconds for browsers acquired from this pool before they are destroyed. Defaults to 600 seconds. Minimum 10, maximum 259200 (72 hours).
viewport
object
No
Browser viewport used for newly-warmed browsers in this pool.
Number of browsers currently acquired from the pool
available_count
number | null
No
Number of browsers currently available in the pool
browser_pool_config
object | null
No
—
created_at
string | null
No
Timestamp when the browser pool was created
extension_ids
string[] | null
No
Resolved extension IDs attached to the pool, in configured load order. Empty when no extensions are attached. Authoritative for programmatic consumers; the extensions inside browser_pool_config reflect the configured selector (echoed as sent on create).
id
string | null
No
Unique identifier for the browser pool
name
string | null
No
Browser pool name, if set
profile_id
string | null
No
Resolved profile ID the pool is attached to. Omitted when no profile is attached. Authoritative for programmatic consumers; the profile inside browser_pool_config reflects the configured selector (echoed as sent on create).
Number of browsers currently acquired from the pool
available_count
number | null
No
Number of browsers currently available in the pool
browser_pool_config
object | null
No
—
created_at
string | null
No
Timestamp when the browser pool was created
extension_ids
string[] | null
No
Resolved extension IDs attached to the pool, in configured load order. Empty when no extensions are attached. Authoritative for programmatic consumers; the extensions inside browser_pool_config reflect the configured selector (echoed as sent on create).
id
string | null
No
Unique identifier for the browser pool
name
string | null
No
Browser pool name, if set
profile_id
string | null
No
Resolved profile ID the pool is attached to. Omitted when no profile is attached. Authoritative for programmatic consumers; the profile inside browser_pool_config reflects the configured selector (echoed as sent on create).
browserPools.listList browser pools and their sizes.Risk:read
await corsair.kernel.api.browserPools.list({});
Input
Name
Type
Required
Description
limit
number
No
Limit the number of browser pools to return.
offset
number
No
Offset the number of browser pools to return.
query
string
No
Case-insensitive substring match against browser pool name. IDs match by exact value.
name
string
No
Exact-match filter on browser pool name using the database collation. In production, matching is case- and accent-insensitive. During the default-project migration, unscoped requests prefer a concrete default-project browser pool over a legacy unscoped browser pool with the same name.
region
us-east | us-west | eu-west | ap-southeast
No
—
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
Whether to reuse the browser instance or destroy it and create a new one. Defaults to true. A reused browser keeps the configuration it was created with, so it does not pick up pool configuration changes made while it was in use. Release with reuse: false, or flush the pool afterward, to rebuild it with the current configuration. Browsers loaded with an acquire-time profile are always destroyed and replaced, even when reuse is true.
session_id
string
Yes
Browser session ID to release back to the pool
Output
Name
Type
Required
Description
success
true
Yes
Kernel accepted the call; it answers with no body.
update
browserPools.updateChange a browser pool’s size or browser configuration. A change applies to browsers the pool creates afterwards: pass discard_all_idle: true to rebuild idle ones now, and release leased ones with reuse: false. To stop loading a profile, send profile: { id: "" }; browsers already in the pool keep the old profile until rebuilt.Risk:write
If provided, replaces the custom Chrome enterprise policy overrides applied to all browsers in this pool. Empty object clears any previously-set policy. Keys are Chrome enterprise policy names; values must match their expected types. Blocked: kernel-managed policies (extensions, proxy, CDP/automation). See https://chromeenterprise.google/policies/ The serialized JSON payload is capped at 5 MiB.
discard_all_idle
boolean
No
Whether to discard all idle browsers and rebuild them immediately with the new configuration. Defaults to false. Only browsers that are idle when the update runs are rebuilt. A browser that is in use during the update keeps its original configuration, except that its allowlist is reset to the current pool baseline on release or the browser is replaced. For other configuration fields, if it is later released with reuse: true it returns to the pool with that stale configuration until it is discarded (by this flag on a later update, or by flushing the pool).
expected_network
object
No
—
extensions
object[]
No
If provided, replaces the extension list. Empty array clears all previously-selected extensions. Omit this field to leave extensions unchanged.
fill_rate_per_minute
number
No
If provided, replaces the percentage of the pool to fill per minute. The cap is 25 for most organizations but can be raised per-organization, so only the lower bound is enforced here.
headless
boolean
No
If provided, replaces whether browsers launch using a headless image.
kiosk_mode
boolean
No
If provided, replaces whether browsers launch in kiosk mode.
memory
1GiB | 2GiB | 4GiB | 8GiB | 16GiB
No
Memory requested for newly-warmed browsers in this pool: 1GiB, 2GiB, or 4GiB for headless, 8GiB or 16GiB for headful. Existing browsers retain their original allocation. Use discard_all_idle to replace idle browsers.
name
string
No
If provided, replaces the pool name. Empty string is a no-op; the pool name cannot be cleared or reset to empty once assigned.
network
object
No
—
profile
object
No
If provided, replaces the previously-selected profile. To clear the profile reference, send { "id": "" } or { "name": "" }. An empty object (no id or name) is rejected as 400.
proxy_id
string
No
Empty string clears the previously-selected proxy. Omit this field to leave the proxy unchanged.
refresh_on_profile_update
boolean
No
If provided, replaces whether idle browsers are flushed when the profile the pool uses is updated. When the pool’s profile reference is changed (including newly attached) and this field is omitted, it defaults to true. Re-sending the same profile reference leaves this setting unchanged. Clearing the profile also disables this setting. Requires a profile to be set on the pool.
size
number
No
If provided, replaces the number of browsers to maintain in the pool. The maximum size is determined by your organization’s pooled sessions limit (the sum of all pool sizes cannot exceed your limit).
start_replay
boolean
No
If provided, replaces whether replay recording starts automatically for future acquired browsers.
start_url
string
No
If provided, replaces the URL to navigate to when a new browser is warmed into the pool. Empty string clears the previously-set URL. Omit this field to leave it unchanged.
stealth
boolean
No
If provided, replaces whether browsers launch in stealth mode. stealth=true without proxy_id moves a pool on direct egress to Kernel’s default stealth proxy. Cannot be combined with captcha_solver.enabled=false. Deprecated in favor of captcha_solver and proxy_id.
telemetry
object | null
No
—
timeout_seconds
number
No
If provided, replaces the default idle timeout in seconds for browsers acquired from this pool before they are destroyed. Minimum 10, maximum 259200 (72 hours).
viewport
object
No
If provided, replaces the viewport with the new width/height/refresh_rate. Because width and height are required and must be positive, the viewport cannot be cleared; omit this field to leave it unchanged.
Number of browsers currently acquired from the pool
available_count
number | null
No
Number of browsers currently available in the pool
browser_pool_config
object | null
No
—
created_at
string | null
No
Timestamp when the browser pool was created
extension_ids
string[] | null
No
Resolved extension IDs attached to the pool, in configured load order. Empty when no extensions are attached. Authoritative for programmatic consumers; the extensions inside browser_pool_config reflect the configured selector (echoed as sent on create).
id
string | null
No
Unique identifier for the browser pool
name
string | null
No
Browser pool name, if set
profile_id
string | null
No
Resolved profile ID the pool is attached to. Omitted when no profile is attached. Authoritative for programmatic consumers; the profile inside browser_pool_config reflects the configured selector (echoed as sent on create).
browsers.computer.batchRun a list of computer actions (click_mouse, type_text, press_key, scroll, move_mouse, drag_mouse, sleep and more) in order in one call; stops at the first error. Faster than one call per action.Risk:write
Kernel accepted the call; it answers with no body.
computer.captureScreenshot
browsers.computer.captureScreenshotTake a PNG screenshot of the browser screen (optionally a region). Headful browsers only: a headless browser has no window, so this returns an empty screen; there, use page.screenshot() inside browsers.playwright.execute. Returns a file: on Corsair it is saved to Files and comes back with a fileId.Risk:read
The file, base64-encoded. Missing when the file came back as text or contentOmitted is set.
text
string
No
The file as text, in place of contentBase64, when it is a text file (CSV, JSON, Markdown, XML and the like) in UTF-8.
contentType
string
Yes
Media type of the file, for example application/pdf.
byteLength
number
Yes
Size of the file in bytes.
filename
string
No
The file name, when known.
contentOmitted
not_requested | over_limit
No
Set when the content was left out: the caller did not ask for it, or the file was over the limit.
fileId
string
No
Where the host saved the file. Over the limit it comes in place of the content; a host that keeps every file sets it beside the content too. On Corsair it is a Files id (fil_…): pass it to another app wherever that app takes a file URL, or read a text file or PDF with files.read_text.
computer.clickMouse
browsers.computer.clickMouseClick at screen coordinates x, y (left, right or middle button; num_clicks for double-click).Risk:write
Duration to hold the keys down in milliseconds. If omitted or 0, keys are tapped.
hold_keys
string[]
No
Optional modifier keys to hold during the key press sequence.
keys
string[]
Yes
List of key symbols to press. Each item should be a key symbol supported by xdotool (see X11 keysym definitions). Examples include “Return”, “Shift”, “Ctrl”, “Alt”, “F5”. Items in this list could also be combinations, e.g. “Ctrl+t” or “Ctrl+Shift+Tab”. Use X11 names for punctuation in combinations, such as “Ctrl+minus” or “Ctrl+plus”. A literal hyphen is also accepted as an alias, so “Ctrl+-” is normalized to “Ctrl+minus”.
Output
Name
Type
Required
Description
success
true
Yes
Kernel accepted the call; it answers with no body.
computer.readClipboard
browsers.computer.readClipboardRead the text on the browser VM’s clipboard.Risk:read
Kernel accepted the call; it answers with no body.
create
browsers.createStart a cloud Chromium browser and get its session_id, cdp_ws_url and browser_live_view_url. Use headless: true unless a person must watch or take over: headless costs about 8x less ($0.06/h vs $0.48/h) and has no live view. stealth: true (or captcha_solver + proxy) helps on bot-protected sites. Pass profile {name, save_changes: true} to start signed in and keep new cookies; the profile is saved when the browser is deleted. start_url opens a page. A browser bills per second while it is active (a CDP, live view or computer-control connection), drops to unbilled standby after about 10 s idle, and is deleted after timeout_seconds of standby (default 60). It holds one of the org’s concurrent-browser slots until deleted, so call browsers.delete when done.Risk:write
Custom Chrome enterprise policy overrides applied to this browser session. Keys are Chrome enterprise policy names; values must match their expected types. Blocked: kernel-managed policies (extensions, proxy, CDP/automation). See https://chromeenterprise.google/policies/
extensions
object[]
No
List of browser extensions to load into the session. Provide each by id or name.
fingerprint
object
No
—
gpu
boolean
No
If true, enables GPU acceleration for the browser session. Requires Start-Up or Enterprise plan, headless=false, and region=us-east.
headless
boolean
No
If true, launches the browser using a headless image (no VNC/GUI). Defaults to false.
invocation_id
string
No
action invocation ID
kiosk_mode
boolean
No
If true, launches the browser in kiosk mode to hide address bar and tabs in live view.
location
object
No
—
memory
1GiB | 2GiB | 4GiB | 8GiB | 16GiB
No
—
name
string
No
Optional human-readable name for the browser session, used to find it later in the dashboard. Must be unique among active sessions within the project. Can be changed later via PATCH /browsers/{id_or_name}.
network
object
No
—
profile
object
No
—
proxy
object
No
—
proxy_id
string
No
Optional proxy to associate to the browser session. Must reference a proxy in the same project as the browser session. Deprecated in favor of proxy.
region
us-east | us-west | eu-west | ap-southeast
No
—
start_replay
boolean
No
Start replay recording as part of creating this browser session.
start_url
string
No
Optional URL to open when the browser session is created. Navigation is best-effort, so navigation failures do not prevent the session from being created.
stealth
boolean
No
If true, enables the CAPTCHA solver and, when proxy is omitted, uses Kernel’s default stealth proxy. Defaults to false. Equivalent to captcha_solver.enabled=true with proxy.mode=default. Cannot be combined with captcha_solver.enabled=false. Deprecated in favor of captcha_solver and proxy.
tags
Record<string, string>
No
—
telemetry
object | null
No
—
timeout_seconds
number
No
The number of seconds of inactivity before the browser session is terminated. Activity includes CDP connections and live view connections. Defaults to 60 seconds. Minimum allowed is 10 seconds. Maximum allowed is 259200 (72 hours). We check for inactivity every 5 seconds, so the actual timeout behavior you will see is +/- 5 seconds around the specified value.
vaults
object[]
No
Project-scoped vaults to link to the browser session. Links are immutable after creation.
Remote URL for live viewing the browser session. Only available for non-headless browsers.
captcha_solver
object | null
No
—
cdp_ws_url
string | null
No
Websocket URL for Chrome DevTools Protocol connections to the browser session
chrome_policy
object | null
No
Custom Chrome enterprise policy overrides that were applied to this browser session, if any. Echoed back for verification. Keys are Chrome enterprise policy names.
cost
object | null
No
—
created_at
string | null
No
When the browser session was created.
deleted_at
string | null
No
When the browser session was soft-deleted. Only present for deleted sessions.
fingerprint
object | null
No
—
gpu
boolean | null
No
Whether GPU acceleration is enabled for the browser session (only supported for headful sessions).
headless
boolean | null
No
Whether the browser session is running in headless mode.
kiosk_mode
boolean | null
No
Whether the browser session is running in kiosk mode.
leased_at
string | null
No
When the pooled browser session was last leased. Only present for leased pooled sessions.
location
object | null
No
—
memory
string | null
No
—
name
string | null
No
Human-readable name of the browser session, if one was set at creation.
network
object | null
No
—
pool
object | null
No
—
profile
object | null
No
—
profile_save_changes
boolean | null
No
Whether changes made during this browser session are saved back to its profile when the session ends. Omitted when no profile is attached.
proxy
object | null
No
—
proxy_id
string | null
No
ID of the proxy associated with this browser session, if any. Deprecated in favor of proxy.
region
string | null
No
—
replay
object | null
No
—
session_id
string | null
No
Unique identifier for the browser session
start_url
string | null
No
URL the session was most recently asked to navigate to, if any. Recorded for debugging. Navigation is fire-and-forget — the URL is dispatched to the browser without waiting for it to load, and any errors (DNS failure, bad status, timeout) are silently dropped. Captures what was requested, not what the browser actually loaded.
stealth
boolean | null
No
Whether the browser session has CAPTCHA solving enabled, in which case proxy.mode=default uses Kernel’s default stealth proxy. A browser created with only captcha_solver reports true with proxy.mode=direct. Deprecated in favor of captcha_solver and proxy.
tags
Record<string, string> | null
No
—
telemetry
object | null
No
—
timeout_seconds
number | null
No
The number of seconds of inactivity before the browser session is terminated.
usage
object | null
No
—
usage_status
string | null
No
—
vaults
object[] | null
No
Vaults linked when the browser session was created.
video_memory
string | null
No
—
viewport
object | null
No
—
webdriver_ws_url
string | null
No
Websocket URL for WebDriver BiDi connections to the browser session
{ height?: number | null, refresh_rate?: number | null, width?: number | null} | null
curl
browsers.curlSend an HTTP request from inside the browser (its cookies, proxy and network fingerprint) and get status, headers and body back. Handy for an API call that needs the signed-in session.Risk:write
GET | HEAD | POST | PUT | PATCH | DELETE | OPTIONS
No
HTTP method.
response_encoding
utf8 | base64
No
Encoding for the response body. Use base64 for binary content.
timeout_ms
number
No
Request timeout in milliseconds.
url
string
Yes
Target URL (must be http or https).
Output
Name
Type
Required
Description
body
string | null
No
Response body (UTF-8 string or base64 depending on request).
duration_ms
number | null
No
Total request duration in milliseconds.
headers
Record<string, string[]> | null
No
Response headers (multi-value).
status
number | null
No
HTTP status code from target.
delete
browsers.deleteEnd a browser session now and stop its billing; frees its concurrency slot. If it was started with profile save_changes: true, the cookies and storage are saved to the profile first. Cannot be undone: the session, its tabs and its files are gone.Risk:write
Kernel accepted the call; it answers with no body.
fs.downloadDirZip
browsers.fs.downloadDirZipDownload a directory from the browser VM as a zip. Returns a file: on Corsair it is saved to Files and comes back with a fileId.Risk:read
The file, base64-encoded. Missing when the file came back as text or contentOmitted is set.
text
string
No
The file as text, in place of contentBase64, when it is a text file (CSV, JSON, Markdown, XML and the like) in UTF-8.
contentType
string
Yes
Media type of the file, for example application/pdf.
byteLength
number
Yes
Size of the file in bytes.
filename
string
No
The file name, when known.
contentOmitted
not_requested | over_limit
No
Set when the content was left out: the caller did not ask for it, or the file was over the limit.
fileId
string
No
Where the host saved the file. Over the limit it comes in place of the content; a host that keeps every file sets it beside the content too. On Corsair it is a Files id (fil_…): pass it to another app wherever that app takes a file URL, or read a text file or PDF with files.read_text.
fs.fileInfo
browsers.fs.fileInfoGet a file’s or directory’s size, mode and modified time in the browser VM.Risk:read
Kernel accepted the call; it answers with no body.
fs.readFile
browsers.fs.readFileRead a file from the browser VM (for example a download in /tmp/downloads). Returns a file: on Corsair it is saved to Files and comes back with a fileId.Risk:read
The file, base64-encoded. Missing when the file came back as text or contentOmitted is set.
text
string
No
The file as text, in place of contentBase64, when it is a text file (CSV, JSON, Markdown, XML and the like) in UTF-8.
contentType
string
Yes
Media type of the file, for example application/pdf.
byteLength
number
Yes
Size of the file in bytes.
filename
string
No
The file name, when known.
contentOmitted
not_requested | over_limit
No
Set when the content was left out: the caller did not ask for it, or the file was over the limit.
fileId
string
No
Where the host saved the file. Over the limit it comes in place of the content; a host that keeps every file sets it beside the content too. On Corsair it is a Files id (fil_…): pass it to another app wherever that app takes a file URL, or read a text file or PDF with files.read_text.
fs.setFilePermissions
browsers.fs.setFilePermissionsChange a file’s mode or owner in the browser VM.Risk:write
Kernel accepted the call; it answers with no body.
fs.watch.start
browsers.fs.watch.startStart watching a directory in the browser VM for changes; returns a watch_id. Kernel streams the events (not available here), so poll browsers.fs.listFiles instead when you can.Risk:write
When true, includes soft-deleted browser sessions in the lookup.
Output
Name
Type
Required
Description
base_url
string | null
No
Metro-API HTTP base URL for this browser session.
browser_live_view_url
string | null
No
Remote URL for live viewing the browser session. Only available for non-headless browsers.
captcha_solver
object | null
No
—
cdp_ws_url
string | null
No
Websocket URL for Chrome DevTools Protocol connections to the browser session
chrome_policy
object | null
No
Custom Chrome enterprise policy overrides that were applied to this browser session, if any. Echoed back for verification. Keys are Chrome enterprise policy names.
cost
object | null
No
—
created_at
string | null
No
When the browser session was created.
deleted_at
string | null
No
When the browser session was soft-deleted. Only present for deleted sessions.
fingerprint
object | null
No
—
gpu
boolean | null
No
Whether GPU acceleration is enabled for the browser session (only supported for headful sessions).
headless
boolean | null
No
Whether the browser session is running in headless mode.
kiosk_mode
boolean | null
No
Whether the browser session is running in kiosk mode.
leased_at
string | null
No
When the pooled browser session was last leased. Only present for leased pooled sessions.
location
object | null
No
—
memory
string | null
No
—
name
string | null
No
Human-readable name of the browser session, if one was set at creation.
network
object | null
No
—
pool
object | null
No
—
profile
object | null
No
—
profile_save_changes
boolean | null
No
Whether changes made during this browser session are saved back to its profile when the session ends. Omitted when no profile is attached.
proxy
object | null
No
—
proxy_id
string | null
No
ID of the proxy associated with this browser session, if any. Deprecated in favor of proxy.
region
string | null
No
—
replay
object | null
No
—
session_id
string | null
No
Unique identifier for the browser session
start_url
string | null
No
URL the session was most recently asked to navigate to, if any. Recorded for debugging. Navigation is fire-and-forget — the URL is dispatched to the browser without waiting for it to load, and any errors (DNS failure, bad status, timeout) are silently dropped. Captures what was requested, not what the browser actually loaded.
stealth
boolean | null
No
Whether the browser session has CAPTCHA solving enabled, in which case proxy.mode=default uses Kernel’s default stealth proxy. A browser created with only captcha_solver reports true with proxy.mode=direct. Deprecated in favor of captcha_solver and proxy.
tags
Record<string, string> | null
No
—
telemetry
object | null
No
—
timeout_seconds
number | null
No
The number of seconds of inactivity before the browser session is terminated.
usage
object | null
No
—
usage_status
string | null
No
—
vaults
object[] | null
No
Vaults linked when the browser session was created.
video_memory
string | null
No
—
viewport
object | null
No
—
webdriver_ws_url
string | null
No
Websocket URL for WebDriver BiDi connections to the browser session
{ height?: number | null, refresh_rate?: number | null, width?: number | null} | null
list
browsers.listList browser sessions in the project, newest first; filter by status (active, deleted, all), tags or name query. Pages with offset/limit; check has_more.Risk:read
Filter sessions by status. “active” returns only active sessions (default), “deleted” returns only soft-deleted sessions, “all” returns both.
include_deleted
boolean
No
Deprecated: Use status=all instead. When true, includes soft-deleted browser sessions in the results alongside active sessions.
limit
number
No
Maximum number of results to return. Defaults to 20, maximum 100.
offset
number
No
Number of results to skip. Defaults to 0.
query
string
No
Search browsers by name, session ID, profile name or ID, proxy ID, or pool name.
tags
Record<string, string>
No
Filter sessions by tag key-value pairs using deepObject style, e.g. ?tags[team]=backend&tags[env]=staging. Multiple pairs are ANDed: a session must match every supplied pair exactly.
region
us-east | us-west | eu-west | ap-southeast
No
—
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
browsers.liveView.createMake a live view link for a headful browser so a person can watch (scope read) or take over (scope control): log in, solve a captcha, approve a step. The link lasts until the browser is deleted or ttl_seconds passes and cannot be revoked, so treat it as a credential and share it only with the person who needs it. Headless browsers have no live view.Risk:write
Kernel accepted the call; it answers with no body.
playwright.execute
browsers.playwright.executeThe simplest way to act on a page: run Playwright code in the browser’s VM in one call and get its return value as result (plus stdout/stderr and error). The code gets page, context, browser and webmcp, e.g. await page.goto('https://example.com'); return await page.title();. timeout_sec defaults to 60 (max 300). Calls with the same executor name run one at a time in a shared process; pass idempotency_key to make a retry safe.Risk:write
Makes retrying the call safe. A call with a key already used on this browser returns the first call’s response instead of running the code again, waiting for it if it is still running. A call with a key keeps running for 10 seconds after its last caller disconnects, so a retry can collect its response; after that it is cancelled like a call without a key. A response is kept for 10 minutes once the call finishes, and only the most recent responses are kept. A call rejected before the code runs, such as with 400, 409 or 429, does not use up its key. The Kernel SDKs send a key automatically and reuse it when they retry.
code
string
Yes
TypeScript/JavaScript code to execute. The code has access to ‘page’, ‘context’, and ‘browser’ variables. It runs within a function, so you can use a return statement at the end to return a value. This value is returned as the result property in the response. Example: “await page.goto(‘https://example.com’); return await page.title();“
browsers.playwright.executors.deleteStop and remove a named Playwright executor on a browser; a later call with that name starts a fresh one.Risk:destructive · Irreversible
browsers.process.execRun a shell command in the browser VM and wait for it: returns exit code, stdout and stderr (base64). Set timeout_sec for long commands.Risk:write
browsers.replRun JavaScript in a persistent Node.js REPL inside the browser VM; variables persist between calls (reset: true starts fresh). Start with repl.help(). Output comes from repl.write, console or repl.emitImage.Risk:write
Makes retrying the call safe. A call with a key already used on this browser returns the first call’s response instead of running the code again, waiting for it if it is still running. A call with a key keeps running for 10 seconds after its last caller disconnects, so a retry can collect its response; after that it is cancelled like a call without a key. A response is kept for 10 minutes once the call finishes, and only the most recent responses are kept. A call rejected before the code runs, such as with 400, 409 or 429, does not use up its key. The Kernel SDKs send a key automatically and reuse it when they retry.
code
string
Yes
JavaScript evaluated in a persistent Node.js runtime. Top-level bindings persist until the browser VM’s API process exits, the REPL is reset, or the REPL is terminated after a crash or timeout. Static top-level imports are unsupported; use dynamic import(). Expression values are ignored; emit output with repl.write(...), console methods, or repl.emitImage(...). May be empty only when reset is true.
reset
boolean
No
Terminate the current REPL, start a fresh one, and then evaluate code.
timeout_sec
number
No
Maximum execution time in seconds. Default is 60.
Output
Name
Type
Required
Description
content
object[] | null
No
Optional ordered text/image output produced by the execution.
content_truncated
boolean | null
No
True if text or image output was dropped or truncated due to response limits.
duration_ms
number | null
No
Wall-clock execution time in milliseconds.
error
string | null
No
Error message if execution failed.
repl_id
string | null
No
CUID2 identifying the exact state-holding REPL process used for this execution. Stable across calls and Chromium reconnects; changes after an API restart, explicit reset, execution timeout, or REPL crash.
repl_terminated
boolean | null
No
True if the REPL identified by repl_id was terminated by this request. The next request lazily starts a fresh REPL with a new repl_id.
browsers.replays.downloadDownload a replay recording as MP4. Returns a file: on Corsair it is saved to Files and comes back with a fileId. Errors with “still preparing” for a few seconds after the recording stops.Risk:read
The file, base64-encoded. Missing when the file came back as text or contentOmitted is set.
text
string
No
The file as text, in place of contentBase64, when it is a text file (CSV, JSON, Markdown, XML and the like) in UTF-8.
contentType
string
Yes
Media type of the file, for example application/pdf.
byteLength
number
Yes
Size of the file in bytes.
filename
string
No
The file name, when known.
contentOmitted
not_requested | over_limit
No
Set when the content was left out: the caller did not ask for it, or the file was over the limit.
fileId
string
No
Where the host saved the file. Over the limit it comes in place of the content; a host that keeps every file sets it beside the content too. On Corsair it is a Files id (fil_…): pass it to another app wherever that app takes a file URL, or read a text file or PDF with files.read_text.
replays.list
browsers.replays.listList a browser’s replay recordings with their ids and view URLs.Risk:read
Kernel accepted the call; it answers with no body.
saveProfile
browsers.saveProfileSave the running browser’s cookies and storage to a profile now, without ending the session: by id (overwrite) or by name (overwrite or create).Risk:write
Optional, easier-to-reference name for the profile
updated_at
string | null
No
Timestamp when the profile was last updated
telemetry.events
browsers.telemetry.eventsRead a page of telemetry events for a browser (console, network, page, interaction, captcha and more), filtered by category, type and time. Pages with offset; check has_more.Risk:read
Opaque pagination cursor: pass the X-Next-Offset value from the previous response to fetch the next page. When set, paging continues from this cursor and since is ignored, while until still bounds the page. It is not an event’s seq field, so do not derive it from the response body.
since
string
No
Start of the window: an RFC-3339 timestamp, or a duration like 5m meaning that long ago. Defaults to 5m. Ignored when offset is set.
until
string
No
End of the window (exclusive): an RFC-3339 timestamp, or a duration like 5m meaning that long ago.
limit
number
No
Maximum number of events per page. Defaults to 20.
Restrict results to these event categories. Repeat the parameter for multiple values.
type
string[]
No
Restrict results to these event types, such as page_crashed or captcha_challenge_result. Repeat the parameter for multiple values. Combines with category: when both are set an event must match both.
order
string
No
Read direction. asc (default) reads oldest first, starting from since or the offset cursor. desc reads newest first: each request returns one page of up to limit records ending at the offset cursor (or until, or the newest archived event); combining desc with since is rejected with a 400.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
If true, stealth browsers connect directly instead of using the default stealth proxy. Deprecated in favor of proxy.mode.
location
object
No
—
name
string | null
No
Human-readable name for the browser session. Omit to leave unchanged, set to an empty string to clear the name. When set, must be unique among active sessions within the project.
network
object
No
—
profile
object
No
—
proxy
object
No
—
proxy_id
string | null
No
ID of the proxy to use. Omit to leave unchanged, set to empty string to remove proxy. Deprecated in favor of proxy.
start_url
string
No
Optional URL to navigate the browser to after applying this update. When a profile is loaded in the same update, this overrides the profile’s restored tabs. Navigation is best-effort, so failures do not fail the update. Omit or set to an empty string to leave the current page unchanged.
Remote URL for live viewing the browser session. Only available for non-headless browsers.
captcha_solver
object | null
No
—
cdp_ws_url
string | null
No
Websocket URL for Chrome DevTools Protocol connections to the browser session
chrome_policy
object | null
No
Custom Chrome enterprise policy overrides that were applied to this browser session, if any. Echoed back for verification. Keys are Chrome enterprise policy names.
cost
object | null
No
—
created_at
string | null
No
When the browser session was created.
deleted_at
string | null
No
When the browser session was soft-deleted. Only present for deleted sessions.
fingerprint
object | null
No
—
gpu
boolean | null
No
Whether GPU acceleration is enabled for the browser session (only supported for headful sessions).
headless
boolean | null
No
Whether the browser session is running in headless mode.
kiosk_mode
boolean | null
No
Whether the browser session is running in kiosk mode.
leased_at
string | null
No
When the pooled browser session was last leased. Only present for leased pooled sessions.
location
object | null
No
—
memory
string | null
No
—
name
string | null
No
Human-readable name of the browser session, if one was set at creation.
network
object | null
No
—
pool
object | null
No
—
profile
object | null
No
—
profile_save_changes
boolean | null
No
Whether changes made during this browser session are saved back to its profile when the session ends. Omitted when no profile is attached.
proxy
object | null
No
—
proxy_id
string | null
No
ID of the proxy associated with this browser session, if any. Deprecated in favor of proxy.
region
string | null
No
—
replay
object | null
No
—
session_id
string | null
No
Unique identifier for the browser session
start_url
string | null
No
URL the session was most recently asked to navigate to, if any. Recorded for debugging. Navigation is fire-and-forget — the URL is dispatched to the browser without waiting for it to load, and any errors (DNS failure, bad status, timeout) are silently dropped. Captures what was requested, not what the browser actually loaded.
stealth
boolean | null
No
Whether the browser session has CAPTCHA solving enabled, in which case proxy.mode=default uses Kernel’s default stealth proxy. A browser created with only captcha_solver reports true with proxy.mode=direct. Deprecated in favor of captcha_solver and proxy.
tags
Record<string, string> | null
No
—
telemetry
object | null
No
—
timeout_seconds
number | null
No
The number of seconds of inactivity before the browser session is terminated.
usage
object | null
No
—
usage_status
string | null
No
—
vaults
object[] | null
No
Vaults linked when the browser session was created.
video_memory
string | null
No
—
viewport
object | null
No
—
webdriver_ws_url
string | null
No
Websocket URL for WebDriver BiDi connections to the browser session
Atomically replace all existing tools in this namespace with this batch when true.
namespace
string
Yes
—
source
string
Yes
JavaScript expression that evaluates to a non-empty array of custom tool definitions. Limited to 8,000,000 bytes when UTF-8 encoded, so multi-byte characters reduce the allowed character count.
Tool input, limited to 4 MiB after JSON serialization.
timeout_sec
number
No
—
tool_ref
string
Yes
—
input full type
{}
Output
Name
Type
Required
Description
error_text
string | null
No
—
invocation_id
string | null
No
—
output
any | null
No
Untrusted page-provided output. Callers must treat it as potentially malicious input.
status
string | null
No
awaiting_submission means a non-autosubmit declarative form was populated but not submitted. Inspect the form, obtain any required confirmation, then submit through Playwright or computer interaction without invoking the tool again. The other statuses are terminal results.
webmcp.listTools
browsers.webmcp.listToolsList the WebMCP tools that open pages expose (native and custom), each with a tool_ref to call it with browsers.webmcp.invokeTool. Tools vanish when their page navigates away.Risk:read
Pollable analysis after workflow submission is acknowledged. Null when no refresh was submitted.
guidance
string | null | null
No
Short advisory markdown to facilitate navigating this target. Returned even when no configuration reached the target, since knowing what prevented success is useful without a configuration. Not verified against this target. Null when nothing applicable was observed or no notes exist.
recommendation
object | null
No
Recommendation or structured no-recommendation result for a completed analysis. Null before the analysis produces an outcome.
target
object | null
No
—
working_configurations
object[] | null
No
Working configurations for the target, ordered with the recommended configuration first.
workload_outcome
string | null | null
No
How far the workload pass got, when an intent was supplied and a pass ran. A run outcome rather than advice, so it is reported whether or not any guidance could be assembled. Null when no intent was supplied or no pass ran.
Pollable analysis after workflow submission is acknowledged. Null when no refresh was submitted.
guidance
string | null | null
No
Short advisory markdown to facilitate navigating this target. Returned even when no configuration reached the target, since knowing what prevented success is useful without a configuration. Not verified against this target. Null when nothing applicable was observed or no notes exist.
recommendation
object | null
No
Recommendation or structured no-recommendation result for a completed analysis. Null before the analysis produces an outcome.
target
object | null
No
—
working_configurations
object[] | null
No
Working configurations for the target, ordered with the recommended configuration first.
workload_outcome
string | null | null
No
How far the workload pass got, when an intent was supplied and a pass ran. A run outcome rather than advice, so it is reported whether or not any guidance could be assembled. Null when no intent was supplied or no pass ran.
ISO 3166 country codes Kernel may use when returning a proxy configuration. When omitted, Kernel uses its default country selection.
url
string
Yes
Public HTTP(S) URL to look up.
Output
Name
Type
Required
Description
guidance
string | null | null
No
Short advisory markdown to facilitate navigating this target. Returned even when no configuration reached the target, since knowing what prevented success is useful without a configuration. Not verified against this target. Null when nothing applicable was observed or no notes exist.
recommendation
object | null
No
Recommendation or structured no-recommendation result from current knowledge. Only the proxy_restricted code is reported here. Null for every other outcome, including a target analyzed without a working configuration.
target
object | null
No
—
working_configurations
object[] | null
No
Working configurations for the target, ordered with the recommended configuration first.
configRegistry.resolveAsk Kernel to analyse a site and work out the browser and proxy settings that get through its bot protection; poll configRegistry.analyses.get.Risk:write
ISO 3166 country codes Kernel may use when searching for or returning a proxy configuration. Kernel may test a subset of allowed countries. When omitted, Kernel uses its default country selection.
intent
string
No
Plain-language description of the workload you intend to run against this target, in a sentence or two. Requires an https target, because the pass treats any non-HTTPS destination as off-site and will not drive an http one. Kernel uses it to drive the browser further into the site, where it can observe protections that only appear once a session interacts. When this target already has a recommended configuration, the run confirms that one instead of re-deriving the whole matrix, so supplying an intent narrows what can be recommended.
url
string
Yes
Public HTTP(S) URL to refresh.
Output
Name
Type
Required
Description
analysis
object | null
No
Pollable analysis after workflow submission is acknowledged. Null when no refresh was submitted.
guidance
string | null | null
No
Short advisory markdown to facilitate navigating this target. Returned even when no configuration reached the target, since knowing what prevented success is useful without a configuration. Not verified against this target. Null when nothing applicable was observed or no notes exist.
recommendation
object | null
No
Recommendation or structured no-recommendation result for a completed analysis. Null before the analysis produces an outcome.
target
object | null
No
—
working_configurations
object[] | null
No
Working configurations for the target, ordered with the recommended configuration first.
workload_outcome
string | null | null
No
How far the workload pass got, when an intent was supplied and a pass ran. A run outcome rather than advice, so it is reported whether or not any guidance could be assembled. Null when no intent was supplied or no pass ran.
Whether the provider is enabled for credential lookups
name
string
No
Human-readable name for this provider instance. Surrounding whitespace is trimmed and the trimmed value must be non-empty.
priority
number
No
Priority order for credential lookups (lower numbers are checked first)
token
string
No
New service account token (to rotate credentials)
Output
Name
Type
Required
Description
created_at
string | null
No
When the credential provider was created
enabled
boolean | null
No
Whether the provider is enabled for credential lookups
id
string | null
No
Unique identifier for the credential provider
name
string | null
No
Human-readable name for this provider instance
priority
number | null
No
Priority order for credential lookups (lower numbers are checked first)
provider_type
string | null
No
Type of credential provider
updated_at
string | null
No
When the credential provider was last updated
Credentials
create
credentials.createStore login values (username, password, TOTP secret) for a domain so managed auth can sign in by itself. Values are write-only.Risk:write
If set, indicates this credential should be used with the specified SSO provider (e.g., google, github, microsoft). When the target site has a matching SSO button, it will be clicked first before filling credential values on the identity provider’s login page.
totp_algorithm
SHA1 | SHA256 | SHA512
No
HMAC algorithm used to generate TOTP codes. Defaults to SHA1 and is ignored when an otpauth:// URI supplies the algorithm.
totp_digits
number
No
Number of digits in generated TOTP codes. Defaults to 6 and is ignored when an otpauth:// URI supplies the digit count.
totp_period
number
No
TOTP rotation period in seconds. Defaults to 30 and is ignored when an otpauth:// URI supplies the period.
totp_secret
string
No
Accepts a 16-128 character base32-encoded TOTP secret or an otpauth://totp/... URI. The range accepts existing shorter seeds and longer seeds regardless of HMAC algorithm; RFC 6238 recommends unpadded base32 lengths of 32/52/103 for SHA1/SHA256/SHA512. Only URI parameters present override the corresponding explicit TOTP fields. Used for automatic 2FA during login.
values
Record<string, string>
Yes
Field name to value mapping (e.g., username, password)
Output
Name
Type
Required
Description
created_at
string | null
No
When the credential was created
domain
string | null
No
Target domain this credential is for
has_totp_secret
boolean | null
No
Whether this credential has a TOTP secret configured for automatic 2FA
has_values
boolean | null
No
Whether this credential has stored values (email, password, etc.)
id
string | null
No
Unique identifier for the credential
name
string | null
No
Unique name for the credential within the project
sso_provider
string | null | null
No
If set, indicates this credential should be used with the specified SSO provider (e.g., google, github, microsoft). When the target site has a matching SSO button, it will be clicked first before filling credential values on the identity provider’s login page.
totp_algorithm
string | null
No
HMAC algorithm used to generate TOTP codes. Defaults to SHA1 for credentials created before this metadata was stored.
totp_code
string | null
No
Current TOTP code. Only included in create/update responses when totp_secret was just set.
totp_code_expires_at
string | null
No
When the totp_code expires. Only included when totp_code is present.
totp_digits
number | null
No
Number of digits in generated TOTP codes. Defaults to 6 for credentials created before this metadata was stored.
totp_period
number | null
No
TOTP rotation period in seconds. Defaults to 30 for credentials created before this metadata was stored.
updated_at
string | null
No
When the credential was last updated
value_keys
string[] | null
No
The field names stored in this credential’s values (e.g., username, password). Values themselves are never returned. Included on single-credential responses (create, get by id or name, update); omitted from list responses.
delete
credentials.deleteDelete a stored credential. Cannot be undone.Risk:destructive · Irreversible
Whether this credential has a TOTP secret configured for automatic 2FA
has_values
boolean | null
No
Whether this credential has stored values (email, password, etc.)
id
string | null
No
Unique identifier for the credential
name
string | null
No
Unique name for the credential within the project
sso_provider
string | null | null
No
If set, indicates this credential should be used with the specified SSO provider (e.g., google, github, microsoft). When the target site has a matching SSO button, it will be clicked first before filling credential values on the identity provider’s login page.
totp_algorithm
string | null
No
HMAC algorithm used to generate TOTP codes. Defaults to SHA1 for credentials created before this metadata was stored.
totp_code
string | null
No
Current TOTP code. Only included in create/update responses when totp_secret was just set.
totp_code_expires_at
string | null
No
When the totp_code expires. Only included when totp_code is present.
totp_digits
number | null
No
Number of digits in generated TOTP codes. Defaults to 6 for credentials created before this metadata was stored.
totp_period
number | null
No
TOTP rotation period in seconds. Defaults to 30 for credentials created before this metadata was stored.
updated_at
string | null
No
When the credential was last updated
value_keys
string[] | null
No
The field names stored in this credential’s values (e.g., username, password). Values themselves are never returned. Included on single-credential responses (create, get by id or name, update); omitted from list responses.
list
credentials.listList stored credentials (names and domains; values are never returned).Risk:read
await corsair.kernel.api.credentials.list({});
Input
Name
Type
Required
Description
domain
string
No
Filter by domain
limit
number
No
Maximum number of results to return
offset
number
No
Number of results to skip
query
string
No
Case-insensitive substring match against credential name or domain. IDs match by exact value.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
Field names to remove from the credential’s stored values. Removals are applied before values are merged, so a key present in both is kept with its new value.
sso_provider
string | null
No
If set, indicates this credential should be used with the specified SSO provider. Set to empty string or null to remove.
totp_algorithm
SHA1 | SHA256 | SHA512
No
HMAC algorithm used to generate TOTP codes. Requires totp_secret and is ignored when an otpauth:// URI supplies the algorithm.
totp_digits
number
No
Number of digits in generated TOTP codes. Requires totp_secret and is ignored when an otpauth:// URI supplies the digit count.
totp_period
number
No
TOTP rotation period in seconds. Requires totp_secret and is ignored when an otpauth:// URI supplies the period.
totp_secret
string
No
Accepts a 16-128 character base32-encoded TOTP secret or an otpauth://totp/... URI. Only URI parameters present override the corresponding explicit TOTP fields. When rotating a raw secret, omitted fields preserve their existing values; a new URI defaults unspecified fields to SHA1/6/30. Set to empty string to remove the secret and its metadata.
values
Record<string, string>
No
Field name to value mapping. Values are merged with existing values (new keys added, existing keys overwritten).
Output
Name
Type
Required
Description
created_at
string | null
No
When the credential was created
domain
string | null
No
Target domain this credential is for
has_totp_secret
boolean | null
No
Whether this credential has a TOTP secret configured for automatic 2FA
has_values
boolean | null
No
Whether this credential has stored values (email, password, etc.)
id
string | null
No
Unique identifier for the credential
name
string | null
No
Unique name for the credential within the project
sso_provider
string | null | null
No
If set, indicates this credential should be used with the specified SSO provider (e.g., google, github, microsoft). When the target site has a matching SSO button, it will be clicked first before filling credential values on the identity provider’s login page.
totp_algorithm
string | null
No
HMAC algorithm used to generate TOTP codes. Defaults to SHA1 for credentials created before this metadata was stored.
totp_code
string | null
No
Current TOTP code. Only included in create/update responses when totp_secret was just set.
totp_code_expires_at
string | null
No
When the totp_code expires. Only included when totp_code is present.
totp_digits
number | null
No
Number of digits in generated TOTP codes. Defaults to 6 for credentials created before this metadata was stored.
totp_period
number | null
No
TOTP rotation period in seconds. Defaults to 30 for credentials created before this metadata was stored.
updated_at
string | null
No
When the credential was last updated
value_keys
string[] | null
No
The field names stored in this credential’s values (e.g., username, password). Values themselves are never returned. Included on single-credential responses (create, get by id or name, update); omitted from list responses.
Deployments
create
deployments.createDeploy a Kernel app from a zip (URL or base64) with its entrypoint, or from a source repository.Risk:write
await corsair.kernel.api.deployments.create({});
Input
Name
Type
Required
Description
entrypoint_rel_path
string
No
Relative path to the entrypoint of the application
env_vars
Record<string, string>
No
Map of environment variables to set for the deployed application. Each key-value pair represents an environment variable.
file
string | object | custom
No
ZIP file containing the application source directory
force
boolean
No
Allow overwriting an existing app version
region
aws.us-east-1a
No
Region for deployment. Currently we only support “aws.us-east-1a”
Environment variables configured for this deployment. Values are redacted for API key, OAuth, and managed-auth callers, which receive every key with an empty string value. Only dashboard sessions receive the actual values.
id
string | null
No
Unique identifier for the deployment
region
string | null
No
Deployment region code
source_checksum
string | null
No
Hex-encoded SHA-256 checksum of the source archive. For file uploads, this hashes the uploaded archive; for GitHub-sourced deployments, this hashes the GitHub archive downloaded by the API. Omitted for deployments created before this field was recorded.
source_path
string | null
No
For GitHub-sourced deployments, the subpath within the repository that was used as the deploy root. Omitted when the repo root was used or for file uploads.
source_ref
string | null
No
For GitHub-sourced deployments, the git ref as requested at deploy time (branch, tag, or commit SHA — not resolved to a commit). Omitted for file uploads.
source_type
string | null
No
Origin of the deployed source code. This is read-only response provenance; file indicates an uploaded archive and github indicates a repository fetched by the API.
source_url
string | null
No
For GitHub-sourced deployments, the repository URL that was fetched. Omitted for file uploads.
status
string | null
No
Current status of the deployment
status_reason
string | null
No
Status reason
updated_at
string | null | null
No
Timestamp when the deployment was last updated
delete
deployments.deleteDelete an app deployment. Cannot be undone.Risk:destructive · Irreversible
Environment variables configured for this deployment. Values are redacted for API key, OAuth, and managed-auth callers, which receive every key with an empty string value. Only dashboard sessions receive the actual values.
id
string | null
No
Unique identifier for the deployment
region
string | null
No
Deployment region code
source_checksum
string | null
No
Hex-encoded SHA-256 checksum of the source archive. For file uploads, this hashes the uploaded archive; for GitHub-sourced deployments, this hashes the GitHub archive downloaded by the API. Omitted for deployments created before this field was recorded.
source_path
string | null
No
For GitHub-sourced deployments, the subpath within the repository that was used as the deploy root. Omitted when the repo root was used or for file uploads.
source_ref
string | null
No
For GitHub-sourced deployments, the git ref as requested at deploy time (branch, tag, or commit SHA — not resolved to a commit). Omitted for file uploads.
source_type
string | null
No
Origin of the deployed source code. This is read-only response provenance; file indicates an uploaded archive and github indicates a repository fetched by the API.
source_url
string | null
No
For GitHub-sourced deployments, the repository URL that was fetched. Omitted for file uploads.
status
string | null
No
Current status of the deployment
status_reason
string | null
No
Status reason
updated_at
string | null | null
No
Timestamp when the deployment was last updated
list
deployments.listList app deployments.Risk:read
await corsair.kernel.api.deployments.list({});
Input
Name
Type
Required
Description
app_name
string
No
Filter results by application name.
app_version
string
No
Filter results by application version. Requires app_name to be set.
limit
number
No
Limit the number of deployments to return.
offset
number
No
Offset the number of deployments to return.
query
string
No
Search deployments by ID or app name.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
The file, base64-encoded. Missing when the file came back as text or contentOmitted is set.
text
string
No
The file as text, in place of contentBase64, when it is a text file (CSV, JSON, Markdown, XML and the like) in UTF-8.
contentType
string
Yes
Media type of the file, for example application/pdf.
byteLength
number
Yes
Size of the file in bytes.
filename
string
No
The file name, when known.
contentOmitted
not_requested | over_limit
No
Set when the content was left out: the caller did not ask for it, or the file was over the limit.
fileId
string
No
Where the host saved the file. Over the limit it comes in place of the content; a host that keeps every file sets it beside the content too. On Corsair it is a Files id (fil_…): pass it to another app wherever that app takes a file URL, or read a text file or PDF with files.read_text.
downloadFromChromeStore
extensions.downloadFromChromeStoreDownload an extension from the Chrome Web Store as an unpacked zip, ready for extensions.upload. Returns a file.Risk:read
Target operating system for the extension package. Defaults to linux.
Output
Name
Type
Required
Description
contentBase64
string
No
The file, base64-encoded. Missing when the file came back as text or contentOmitted is set.
text
string
No
The file as text, in place of contentBase64, when it is a text file (CSV, JSON, Markdown, XML and the like) in UTF-8.
contentType
string
Yes
Media type of the file, for example application/pdf.
byteLength
number
Yes
Size of the file in bytes.
filename
string
No
The file name, when known.
contentOmitted
not_requested | over_limit
No
Set when the content was left out: the caller did not ask for it, or the file was over the limit.
fileId
string
No
Where the host saved the file. Over the limit it comes in place of the content; a host that keeps every file sets it beside the content too. On Corsair it is a Files id (fil_…): pass it to another app wherever that app takes a file URL, or read a text file or PDF with files.read_text.
get
extensions.getGet an uploaded extension’s metadata.Risk:read
SHA-256 checksum, encoded as lowercase hexadecimal, of the exact uploaded extension archive bytes. This is not a normalized checksum of the extension contents; archive metadata, file ordering, and compression can change the checksum for otherwise identical contents. Omitted for legacy rows and server-repackaged Chrome Web Store extensions.
created_at
string | null
No
Timestamp when the extension was created
id
string | null
No
Unique identifier for the extension
last_used_at
string | null | null
No
Timestamp when the extension was last used
name
string | null | null
No
Optional, easier-to-reference name for the extension. Must be unique within the project.
Case-insensitive substring match against extension name. IDs match by exact value.
name
string
No
Exact-match filter on extension name using the database collation. In production, matching is case- and accent-insensitive. During the default-project migration, unscoped requests prefer a concrete default-project extension over a legacy unscoped extension with the same name.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
SHA-256 checksum, encoded as lowercase hexadecimal, of the exact uploaded extension archive bytes. This is not a normalized checksum of the extension contents; archive metadata, file ordering, and compression can change the checksum for otherwise identical contents. Omitted for legacy rows and server-repackaged Chrome Web Store extensions.
created_at
string | null
No
Timestamp when the extension was created
id
string | null
No
Unique identifier for the extension
last_used_at
string | null | null
No
Timestamp when the extension was last used
name
string | null | null
No
Optional, easier-to-reference name for the extension. Must be unique within the project.
size_bytes
number | null
No
Size of the extension archive in bytes
Invocations
create
invocations.createRun an action of a deployed Kernel app (billed per second). async: true returns at once; poll invocations.get for the result.Risk:destructive · Irreversible
If true, invoke asynchronously. When set, the API responds 202 Accepted with status “queued”.
async_timeout_seconds
number
No
Timeout in seconds for async invocations (min 10, max 3600). Only applies when async is true.
payload
string
No
Input data for the action, sent as a JSON string.
version
string
Yes
Version of the application
Output
Name
Type
Required
Description
action_name
string | null
No
Name of the action invoked
id
string | null
No
ID of the invocation
output
string | null
No
The action result or detailed failure output. Often a JSON-encoded value, but failures may contain plain text. May contain sensitive application data.
status
string | null
No
Status of the invocation
status_reason
string | null
No
A nonempty, customer-safe summary of the recorded failure output, always present when status is failed and omitted otherwise. Recognized messages receive a specific summary; other failures receive a generic summary. Message matching does not establish whether the failure originated in the platform or action code. Does not include raw action output or internal error details. Human-readable text, not a stable identifier for retry logic.
deleteBrowsers
invocations.deleteBrowsersDelete every browser an invocation started. Cannot be undone.Risk:destructive · Irreversible
RFC 3339 Nanoseconds timestamp when the invocation finished (null if still running)
id
string | null
No
ID of the invocation
output
string | null
No
The action result or detailed failure output. Often a JSON-encoded value, but failures may contain plain text. May contain sensitive application data.
payload
string | null
No
Payload provided to the invocation. This is a string that can be parsed as JSON.
started_at
string | null
No
RFC 3339 Nanoseconds timestamp when the invocation started
status
string | null
No
Status of the invocation
status_reason
string | null
No
A nonempty, customer-safe summary of the recorded failure output, always present when status is failed and omitted otherwise, including in the first failed invocation_state event. Recognized messages receive a specific summary; other failures receive a generic summary. Message matching does not establish whether the failure originated in the platform or action code. Does not include raw action output or internal error details. Available for historical invocations as well. Human-readable text, not a stable identifier for retry logic.
version
string | null
No
Version label for the application
list
invocations.listList app action invocations, filtered by app, action, status or time.Risk:read
await corsair.kernel.api.invocations.list({});
Input
Name
Type
Required
Description
app_name
string
No
Filter results by application name.
version
string
No
Filter results by application version.
action_name
string
No
Filter results by action name.
deployment_id
string
No
Filter results by deployment ID.
status
queued | running | succeeded | failed
No
Filter results by invocation status.
since
string
No
Show invocations that have started since the given time (RFC timestamps or durations like 5m).
limit
number
No
Limit the number of invocations to return.
offset
number
No
Offset the number of invocations to return.
query
string
No
Search invocations by ID, app name, or action name.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
Updated output of the invocation rendered as JSON string.
status
succeeded | failed
Yes
New status for the invocation.
Output
Name
Type
Required
Description
action_name
string | null
No
Name of the action invoked
app_name
string | null
No
Name of the application
finished_at
string | null | null
No
RFC 3339 Nanoseconds timestamp when the invocation finished (null if still running)
id
string | null
No
ID of the invocation
output
string | null
No
The action result or detailed failure output. Often a JSON-encoded value, but failures may contain plain text. May contain sensitive application data.
payload
string | null
No
Payload provided to the invocation. This is a string that can be parsed as JSON.
started_at
string | null
No
RFC 3339 Nanoseconds timestamp when the invocation started
status
string | null
No
Status of the invocation
status_reason
string | null
No
A nonempty, customer-safe summary of the recorded failure output, always present when status is failed and omitted otherwise, including in the first failed invocation_state event. Recognized messages receive a specific summary; other failures receive a generic summary. Message matching does not establish whether the failure originated in the platform or action code. Does not include raw action output or internal error details. Available for historical invocations as well. Human-readable text, not a stable identifier for retry logic.
version
string | null
No
Version label for the application
Organization
entitlements.get
organization.entitlements.getGet what the org’s plan includes (features and quotas).Risk:read
The organization’s current non-deleted managed auth connections, counted org-wide across every project. Compare against max_auth_connections to show remaining capacity before a create is rejected with 403 insufficient_plan.
concurrent_sessions_available
number | null | null
No
Number of concurrent browser slots currently available to the organization. This is the effective concurrency limit minus active on-demand sessions and browser pool reservations, floored at zero. Null when usage cannot be read.
concurrent_sessions_used
number | null | null
No
Current organization-wide concurrent browser usage, including active on-demand sessions and browser pool reservations. Null when usage cannot be read.
default_project_max_concurrent_sessions
number | null | null
No
Default maximum concurrent browsers applied to every project that has no explicit per-project override. Null means no org-level default, so such projects are uncapped (only the org-wide limit applies). Applies to existing and newly created projects.
max_auth_connections
number | null | null
No
Maximum managed auth connections the organization’s plan allows. Null means unlimited. Counted org-wide, so it cannot be multiplied across projects.
max_concurrent_sessions
number | null
No
The organization’s effective concurrency limit — the maximum browsers running at once, covering both on-demand sessions and browser pool reservations — from its plan or an override. Read-only and shared across all projects in the org; a per-project default cannot exceed it.
max_vaults
number | null | null
No
Maximum non-deleted vaults allowed org-wide across all projects. Null means unlimited.
min_health_check_interval_seconds
number | null
No
Smallest health_check_interval the organization’s plan accepts on a managed auth connection. Requests below this are rejected with 400. Existing connections stored below the floor are grandfathered until edited.
vaults_used
number | null
No
Current non-deleted vault count across all projects in the organization.
limits.update
organization.limits.updateSet the default concurrent-browser cap for projects without their own.Risk:write
Default maximum concurrent browsers for projects without an explicit override. Set to 0 to remove the default; omit to leave unchanged. Cannot exceed the organization’s concurrency limit.
Output
Name
Type
Required
Description
auth_connections_used
number | null
No
The organization’s current non-deleted managed auth connections, counted org-wide across every project. Compare against max_auth_connections to show remaining capacity before a create is rejected with 403 insufficient_plan.
concurrent_sessions_available
number | null | null
No
Number of concurrent browser slots currently available to the organization. This is the effective concurrency limit minus active on-demand sessions and browser pool reservations, floored at zero. Null when usage cannot be read.
concurrent_sessions_used
number | null | null
No
Current organization-wide concurrent browser usage, including active on-demand sessions and browser pool reservations. Null when usage cannot be read.
default_project_max_concurrent_sessions
number | null | null
No
Default maximum concurrent browsers applied to every project that has no explicit per-project override. Null means no org-level default, so such projects are uncapped (only the org-wide limit applies). Applies to existing and newly created projects.
max_auth_connections
number | null | null
No
Maximum managed auth connections the organization’s plan allows. Null means unlimited. Counted org-wide, so it cannot be multiplied across projects.
max_concurrent_sessions
number | null
No
The organization’s effective concurrency limit — the maximum browsers running at once, covering both on-demand sessions and browser pool reservations — from its plan or an override. Read-only and shared across all projects in the org; a per-project default cannot exceed it.
max_vaults
number | null | null
No
Maximum non-deleted vaults allowed org-wide across all projects. Null means unlimited.
min_health_check_interval_seconds
number | null
No
Smallest health_check_interval the organization’s plan accepts on a managed auth connection. Requests below this are rejected with 400. Existing connections stored below the floor are grandfathered until edited.
vaults_used
number | null
No
Current non-deleted vault count across all projects in the organization.
Profiles
create
profiles.createCreate an empty browser profile (saved cookies and storage) to start browsers signed in. Fill it by starting a browser with profile {name, save_changes: true}, logging in, then deleting the browser.Risk:write
Optional name of the profile. Must be unique within the logical project; during the default-project migration, unscoped profiles and profiles in the org default project are treated as the same project.
Output
Name
Type
Required
Description
created_at
string | null
No
Timestamp when the profile was created
id
string | null
No
Unique identifier for the profile
last_used_at
string | null
No
Timestamp when the profile was last used
name
string | null | null
No
Optional, easier-to-reference name for the profile
updated_at
string | null
No
Timestamp when the profile was last updated
delete
profiles.deleteDelete a browser profile and its saved cookies and storage. Cannot be undone.Risk:destructive · Irreversible
Kernel accepted the call; it answers with no body.
download
profiles.downloadDownload a profile’s saved state as a tar.zst (or tar) archive. Returns a file. Holds session cookies: treat it as a credential.Risk:read
Response format for current profile archives. Legacy profiles are always returned as JSON.
Output
Name
Type
Required
Description
contentBase64
string
No
The file, base64-encoded. Missing when the file came back as text or contentOmitted is set.
text
string
No
The file as text, in place of contentBase64, when it is a text file (CSV, JSON, Markdown, XML and the like) in UTF-8.
contentType
string
Yes
Media type of the file, for example application/pdf.
byteLength
number
Yes
Size of the file in bytes.
filename
string
No
The file name, when known.
contentOmitted
not_requested | over_limit
No
Set when the content was left out: the caller did not ask for it, or the file was over the limit.
fileId
string
No
Where the host saved the file. Over the limit it comes in place of the content; a host that keeps every file sets it beside the content too. On Corsair it is a Files id (fil_…): pass it to another app wherever that app takes a file URL, or read a text file or PDF with files.read_text.
get
profiles.getGet a browser profile by id or name.Risk:read
Optional, easier-to-reference name for the profile
updated_at
string | null
No
Timestamp when the profile was last updated
list
profiles.listList saved browser profiles (name, id, last used).Risk:read
await corsair.kernel.api.profiles.list({});
Input
Name
Type
Required
Description
limit
number
No
Limit the number of profiles to return.
offset
number
No
Offset the number of profiles to return.
query
string
No
Case-insensitive substring match against profile name or ID.
name
string
No
Exact-match filter on profile name using the database collation. In production, matching is case- and accent-insensitive. During the default-project migration, unscoped requests prefer a concrete default-project profile over a legacy unscoped profile with the same name.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
New profile name. Must be unique within the logical project; during the default-project migration, unscoped profiles and profiles in the org default project are treated as the same project.
Output
Name
Type
Required
Description
created_at
string | null
No
Timestamp when the profile was created
id
string | null
No
Unique identifier for the profile
last_used_at
string | null
No
Timestamp when the profile was last used
name
string | null | null
No
Optional, easier-to-reference name for the profile
updated_at
string | null
No
Timestamp when the profile was last updated
Projects
create
projects.createCreate a project to keep browsers, profiles and keys separate within the org.Risk:write
Maximum concurrent app invocations for this project. Null means no project-level cap.
max_concurrent_sessions
number | null | null
No
Maximum concurrent browsers for this project, covering both on-demand sessions (browsers.create()) and browser pool reservations. Null means no project-level cap.
max_pooled_sessions
number | null | null
No
Deprecated: pooled browsers now count toward max_concurrent_sessions. Always null.
limits.update
projects.limits.updateSet a project’s concurrent-browser cap.Risk:write
Maximum concurrent app invocations for this project. Set to 0 to remove the cap; omit to leave unchanged.
max_concurrent_sessions
number | null
No
Maximum concurrent browsers for this project, covering both on-demand sessions and browser pool reservations. Set to 0 to remove the cap; omit to leave unchanged.
max_pooled_sessions
number | null
No
Deprecated: pooled browsers now count toward max_concurrent_sessions. Requests that set this field are rejected with a 400.
Output
Name
Type
Required
Description
max_concurrent_invocations
number | null | null
No
Maximum concurrent app invocations for this project. Null means no project-level cap.
max_concurrent_sessions
number | null | null
No
Maximum concurrent browsers for this project, covering both on-demand sessions (browsers.create()) and browser pool reservations. Null means no project-level cap.
max_pooled_sessions
number | null | null
No
Deprecated: pooled browsers now count toward max_concurrent_sessions. Always null.
list
projects.listList the org’s projects.Risk:read
await corsair.kernel.api.projects.list({});
Input
Name
Type
Required
Description
limit
number
No
Maximum number of results to return
offset
number
No
Number of results to skip
query
string
No
Case-insensitive substring match against project name
name
string
No
Exact-match filter on project name using the database collation. In production, matching is case- and accent-insensitive.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
Proxy ID. Name-based addressing is supported only for retrieval.
url
string
No
An optional URL to test reachability against. If provided, the proxy check will test connectivity to this URL instead of the default test URLs. Only HTTP and HTTPS schemes are allowed, and the URL must resolve to a public IP address. For ISP proxies, the exit IP is stable, so a successful check reliably indicates that subsequent browser sessions will reach the target site with the same IP. For residential and mobile proxies, the exit node changes between requests, so a successful check validates proxy configuration but does not guarantee that a subsequent browser session will use the same exit IP or reach the same site — it is useful for verifying credentials and connectivity, not for predicting site-specific behavior. When provided, the check result does not update the proxy’s health status, since a failure may indicate a problem with the target site rather than the proxy itself.
Output
Name
Type
Required
Description
bypass_hosts
string[] | null
No
Hostnames that should bypass the parent proxy and connect directly.
config
object | null
No
Configuration specific to the selected proxy type.
id
string | null
No
—
ip_address
string | null
No
IP address that the proxy uses when making requests.
last_checked
string | null
No
Timestamp of the last health check performed on this proxy.
name
string | null
No
Readable name of the proxy.
protocol
string | null
No
Protocol to use for the proxy connection.
status
string | null
No
Current health status of the proxy.
type
string | null
No
Proxy type to use. In terms of quality for avoiding bot-detection, from best to worst: mobile > residential > isp. datacenter is deprecated; use isp instead. Existing datacenter proxies remain supported.
Hostnames that should bypass the parent proxy and connect directly.
config
object
No
Configuration specific to the selected proxy type.
name
string
No
Readable name of the proxy.
protocol
http | https
No
Protocol to use for the proxy connection.
type
datacenter | isp | residential | mobile | custom
Yes
Proxy type to use. In terms of quality for avoiding bot-detection, from best to worst: mobile > residential > isp. datacenter is deprecated; use isp instead. Datacenter proxy creation remains supported for backward compatibility.
Hostnames that should bypass the parent proxy and connect directly.
config
object | null
No
Configuration specific to the selected proxy type.
id
string | null
No
—
ip_address
string | null
No
IP address that the proxy uses when making requests.
last_checked
string | null
No
Timestamp of the last health check performed on this proxy.
name
string | null
No
Readable name of the proxy.
protocol
string | null
No
Protocol to use for the proxy connection.
status
string | null
No
Current health status of the proxy.
type
string | null
No
Proxy type to use. In terms of quality for avoiding bot-detection, from best to worst: mobile > residential > isp. datacenter is deprecated; use isp instead. Existing datacenter proxies remain supported.
Hostnames that should bypass the parent proxy and connect directly.
config
object | null
No
Configuration specific to the selected proxy type.
id
string | null
No
—
ip_address
string | null
No
IP address that the proxy uses when making requests.
last_checked
string | null
No
Timestamp of the last health check performed on this proxy.
name
string | null
No
Readable name of the proxy.
protocol
string | null
No
Protocol to use for the proxy connection.
status
string | null
No
Current health status of the proxy.
type
string | null
No
Proxy type to use. In terms of quality for avoiding bot-detection, from best to worst: mobile > residential > isp. datacenter is deprecated; use isp instead. Existing datacenter proxies remain supported.
Case-insensitive substring match against proxy name, host, or IP address. IDs match by exact value.
name
string
No
Exact-match filter on proxy name using the database collation. In production, matching is case- and accent-insensitive. Names are not required to be unique, so multiple proxies may match.
ip_address
string
No
Exact match on the primary exit IP of ISP proxies. Must be an IPv4 address. Only ISP proxies are matched: the ip_address shown on other proxy types is a one-time observation, not a stable identity.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
Proxy ID. Name-based addressing is supported only for retrieval.
config
object
No
Type-specific changes. Only ISP proxies accept a config update.
name
string
No
New proxy name. Proxy names are trimmed and length-checked only; duplicates are allowed because proxies are updated by ID, not by name.
config full type
{ ip_address: string}
Output
Name
Type
Required
Description
bypass_hosts
string[] | null
No
Hostnames that should bypass the parent proxy and connect directly.
config
object | null
No
Configuration specific to the selected proxy type.
id
string | null
No
—
ip_address
string | null
No
IP address that the proxy uses when making requests.
last_checked
string | null
No
Timestamp of the last health check performed on this proxy.
name
string | null
No
Readable name of the proxy.
protocol
string | null
No
Protocol to use for the proxy connection.
status
string | null
No
Current health status of the proxy.
type
string | null
No
Proxy type to use. In terms of quality for avoiding bot-detection, from best to worst: mobile > residential > isp. datacenter is deprecated; use isp instead. Existing datacenter proxies remain supported.
Makes retrying the create safe. A retry with the same key and the same request returns the record the first call created, with status 200, instead of creating another. Reusing a key with a different request returns 422. Keys are scoped to the project and do not expire. The Kernel SDKs send a key automatically and reuse it when they retry.
data
object
Yes
Report data. It must match the schema of the report type’s version and encode to at most 16 KB of JSON.
occurred_at
string
No
When the reported event happened. Defaults to the time the report is submitted. Must be no more than 5 minutes in the future or 30 days in the past.
schema_version
number
No
Version of the report type’s schema to validate data against. Defaults to the type’s latest version, which the response returns.
session_id
string
No
ID of the browser the report was observed in. It must belong to the task’s project; deleted browsers are accepted. The browser is linked to the task if it is not already.
task_id
string
Yes
ID of the task to report against.
type
string
Yes
Report type name. List report types to see the available types and their schemas.
data full type
{}
Output
Name
Type
Required
Description
created_at
string | null
No
When the report was submitted
data
object | null
No
Report data, matching the report type’s schema
id
string | null
No
Unique identifier for the report
occurred_at
string | null
No
When the reported event happened
schema_version
number | null
No
Version of the report type’s schema that data was validated against
session_id
string | null | null
No
ID of the browser the report was observed in, if any
What the report type records and when to submit it
latest_version
number | null
No
Highest version that is not retired. A report submitted without a schema_version uses this version. It is 0 when every version is retired.
schema
object | null
No
JSON Schema (draft 2020-12) of the latest version, which a report’s data must match. Its field descriptions and examples describe what to send. Null when every version is retired.
type
string | null
No
Report type name, such as task.result
versions
object[] | null
No
Every version of the type, oldest first, including retired versions.
Maximum number of search results to fetch when result_ids is omitted, starting from rank 1. Mutually exclusive with result_ids.
result_ids
string[]
No
Kernel-generated IDs from the referenced retained search, in desired response order. They are not provider-standard IDs. Mutually exclusive with limit.
search.createSearch the web through one of Kernel’s search providers and get ranked results (titles, URLs, snippets), kept for search.contents.fetch.Risk:read
Optional portable content retrieval. Pass true for defaults or an options object. Omission never starts Kernel browser work; provider-supplied content is still returned when available, including when requested through native options. Both inline and deferred retrieval use the same options schema.
country
string
No
ISO 3166-1 alpha-2 search locale preference.
end_date
string
No
Inclusive publication-date upper bound; must not precede start_date. If recency is also supplied, recency takes precedence with a warning. Unsupported or approximated filtering is reported, or rejected under strict_params.
exclude_domains
string[]
No
Hostname exclusions, with the same best-effort/strict behavior as include_domains. Provider-specific combinations that cannot be represented are reported via warnings or rejected in strict mode.
include_domains
string[]
No
Hostname inclusion preference, matching a hostname and its subdomains. Empty means unrestricted. Translated, emulated, or dropped with a warning according to provider capability unless strict_params is true. Native boost modes remain advisory and are identified in warnings.
include_raw
boolean
No
Include untouched per-result payloads and the full serving-provider response in raw fields. Off by default; native top-level outputs such as answer remain available without it.
language
string
No
BCP 47 search language preference.
max_results
number
No
Requested result count from 1 through 100. The effective count is clamped to the serving provider’s cap with a warning. Effective native counts are the lower of this limit and supplied provider-native count aliases. Strict mode rejects unsupported counts.
query
string
Yes
Primary search query. A provider-native multi-query option applies only to that provider; other providers in a fallback chain receive this query.
recency
hour | day | week | month | year
No
Relative search window. Takes precedence over start_date/end_date with a warning if both are set. Provider-native recency behavior is retained, including documented hour-to-day widening. Unsupported filters are rejected only in strict mode.
safe_search
off | moderate | strict
No
Optional safety preference. Omit to use provider defaults. Unsupported values are dropped with a warning unless strict_params is true. A search filter is not an authorization boundary.
start_date
string
No
Inclusive publication-date lower bound. If recency is also supplied, recency takes precedence with a warning. Provider date semantics, precision, and unsupported filters are reported; unknown source dates are not fabricated or universally post-filtered.
strategy
object
No
—
strict_params
boolean
No
When false, unsupported portable parameters are omitted and approximations are described in warnings. When true, every supplied portable parameter must be honored exactly. Requests that cannot be served with those parameters are rejected. This does not guarantee identical rankings or document timestamps across indexes. Authentication and project isolation are always enforced.
timeout_ms
number
No
Overall deadline across search attempts and inline retrieval. No new attempt starts after the deadline. Completed search results survive inline retrieval timeouts.
Concrete serving provider, never auto or fallback.
query
string | null
No
Echo of the query. Native multi-query inputs are visible in the selected strategy target and the optional raw response.
raw
any | null
No
Full serving-provider response, including top-level metadata that does not belong to a result. Present only with include_raw=true; untrusted provider data.
Concrete serving provider, never auto or fallback.
query
string | null
No
Echo of the query. Native multi-query inputs are visible in the selected strategy target and the optional raw response.
raw
any | null
No
Full serving-provider response, including top-level metadata that does not belong to a result. Present only with include_raw=true; untrusted provider data.
Makes retrying the create safe. A retry with the same key and the same request returns the record the first call created, with status 200, instead of creating another. Reusing a key with a different request returns 422. Keys are scoped to the project and do not expire. The Kernel SDKs send a key automatically and reuse it when they retry.
description
string
Yes
What the task is meant to accomplish, at most 4 KB.
metadata
Record<string, string>
No
—
session_ids
string[]
No
IDs of browsers to link to the task. Each must belong to the request’s project; deleted browsers can be linked.
started_at
string
No
When work on the task started. Defaults to the time the task is created. Must be no more than 5 minutes in the future or 30 days in the past.
target_website
string
No
Hostname of the website the task targets, without a scheme or path.
Output
Name
Type
Required
Description
created_at
string | null
No
When the task was created
description
string | null
No
What the task is meant to accomplish
id
string | null
No
Unique identifier for the task
metadata
Record<string, string> | null
No
—
session_ids
string[] | null
No
IDs of browsers linked to the task, oldest link first. Linked browsers may since have been deleted.
IDs of browsers to link. Each must belong to the task’s project; deleted browsers can be linked. Browsers already linked are ignored. A task can have at most 100 linked browsers.
Output
Name
Type
Required
Description
created_at
string | null
No
When the task was created
description
string | null
No
What the task is meant to accomplish
id
string | null
No
Unique identifier for the task
metadata
Record<string, string> | null
No
—
session_ids
string[] | null
No
IDs of browsers linked to the task, oldest link first. Linked browsers may since have been deleted.
started_at
string | null
No
When work on the task started
target_website
string | null | null
No
Hostname of the website the task targets
list
tasks.listList tasks, newest first.Risk:read
await corsair.kernel.api.tasks.list({});
Input
Name
Type
Required
Description
limit
number
No
Limit the number of tasks to return.
offset
number
No
Offset the number of tasks to return.
session_id
string
No
Only return tasks linked to this browser ID.
Output
Name
Type
Required
Description
items
object[]
Yes
This page of results.
has_more
boolean
Yes
True when more results follow; pass next_offset as offset to read them.
next_offset
number | null
Yes
Value for offset that reads the next page, or null on the last page.
Base endpoint of the OTLP/HTTP collector, without a signal path. Kernel appends the signal path itself, so pass https://api.honeycomb.io rather than https://api.honeycomb.io/v1/logs. If your provider’s docs give you a signal-specific URL, drop the trailing /v1/logs, /v1/traces, or /v1/metrics — an endpoint that already carries one is rejected. Must be http or https, must resolve to a public address, and must carry no query string or fragment. Examples: https://api.honeycomb.io, https://otlp-gateway-prod-us-east-0.grafana.net/otlp, https://otlp.datadoghq.com (Datadog’s OTLP intake for US1, not its logs intake).
headers
Record<string, string>
No
Headers sent with each export request, typically an ingestion key. Encrypted at rest and returned redacted. Names and values must be valid HTTP header tokens, and the names and values together cannot exceed 8192 bytes. Names are matched case-insensitively and stored canonicalized, so supplying two spellings of one header is rejected.
name
string
Yes
Unique within the organization.
Output
Name
Type
Required
Description
consecutive_failures
number | null
No
Failed deliveries since the last success, as observed by the relay process that wrote the latest outcome. Zero means the most recently recorded outcome succeeded.
created_at
string | null
No
—
description
string | null
No
—
endpoint
string | null
No
OTLP/HTTP endpoint telemetry is sent to.
headers
Record<string, string> | null
No
Headers sent with each export request. Names are returned in canonical form (Authorization, not authorization). Non-dashboard reads return values redacted as empty strings, so the keys are visible but the credentials are not. Dashboard reads return the stored values.
id
string | null
No
—
last_error
string | null
No
Sanitized class of the delivery failure recorded at last_error_at. It is retained after a later success, so its presence does not mean the destination is currently failing. Response bodies, endpoint URLs, credentials, and raw transport errors are never returned.
last_error_at
string | null
No
Timestamp of the most recent failed delivery. It is retained after a later success, so it can predate last_export_at. Read consecutive_failures to tell whether the destination is currently failing.
last_export_at
string | null
No
Timestamp of the most recent successful delivery. Moves only on success.
name
string | null
No
Unique within the organization. Usable in place of the ID when selecting a destination, so it cannot be shaped like an ID.
updated_at
string | null
No
—
destinations.delete
telemetry.destinations.deleteDelete an OTLP telemetry destination. Cannot be undone.Risk:destructive · Irreversible
Failed deliveries since the last success, as observed by the relay process that wrote the latest outcome. Zero means the most recently recorded outcome succeeded.
created_at
string | null
No
—
description
string | null
No
—
endpoint
string | null
No
OTLP/HTTP endpoint telemetry is sent to.
headers
Record<string, string> | null
No
Headers sent with each export request. Names are returned in canonical form (Authorization, not authorization). Non-dashboard reads return values redacted as empty strings, so the keys are visible but the credentials are not. Dashboard reads return the stored values.
id
string | null
No
—
last_error
string | null
No
Sanitized class of the delivery failure recorded at last_error_at. It is retained after a later success, so its presence does not mean the destination is currently failing. Response bodies, endpoint URLs, credentials, and raw transport errors are never returned.
last_error_at
string | null
No
Timestamp of the most recent failed delivery. It is retained after a later success, so it can predate last_export_at. Read consecutive_failures to tell whether the destination is currently failing.
last_export_at
string | null
No
Timestamp of the most recent successful delivery. Moves only on success.
name
string | null
No
Unique within the organization. Usable in place of the ID when selecting a destination, so it cannot be shaped like an ID.
Base endpoint of the OTLP/HTTP collector, without a signal path. Same rules as on create.
headers
Record<string, string | null>
No
Edits stored headers key by key rather than replacing the map. A string value adds or replaces that header, null deletes it, and any key you omit is left as it is. Names are matched case-insensitively, so authorization replaces a stored Authorization rather than adding a second entry. This is the credential rotation path; sessions already exporting pick up the new values without restarting. Names and values must be valid HTTP header tokens, and the names and values together cannot exceed 8192 bytes.
name
string
No
—
Output
Name
Type
Required
Description
consecutive_failures
number | null
No
Failed deliveries since the last success, as observed by the relay process that wrote the latest outcome. Zero means the most recently recorded outcome succeeded.
created_at
string | null
No
—
description
string | null
No
—
endpoint
string | null
No
OTLP/HTTP endpoint telemetry is sent to.
headers
Record<string, string> | null
No
Headers sent with each export request. Names are returned in canonical form (Authorization, not authorization). Non-dashboard reads return values redacted as empty strings, so the keys are visible but the credentials are not. Dashboard reads return the stored values.
id
string | null
No
—
last_error
string | null
No
Sanitized class of the delivery failure recorded at last_error_at. It is retained after a later success, so its presence does not mean the destination is currently failing. Response bodies, endpoint URLs, credentials, and raw transport errors are never returned.
last_error_at
string | null
No
Timestamp of the most recent failed delivery. It is retained after a later success, so it can predate last_export_at. Read consecutive_failures to tell whether the destination is currently failing.
last_export_at
string | null
No
Timestamp of the most recent successful delivery. Moves only on success.
name
string | null
No
Unique within the organization. Usable in place of the ID when selecting a destination, so it cannot be shaped like an ID.
updated_at
string | null
No
—
Vault Provider Configs
create
vaultProviderConfigs.createRegister a vault provider configuration (card or payment provider settings shared by the org’s projects).Risk:write
When true, also finds deleted items, and deleted vaults by ID only. key then also accepts an item ID, which takes precedence over a key with the same value; the lookup order is the item ID, then the live item with that key, then the most recently deleted one. wait is ignored for a deleted item or vault.
wait
number
No
Long-poll for new events for up to this many seconds.
Hold for up to this many seconds while the item is pending authorization, approval, or credential collection. Return the current item when ready or when the wait elapses. This does not wait for edits to an already-ready credential; poll GET without wait and compare version to observe changes after collect. Managed auth credentials are created ready, so wait returns immediately.
vaults.items.performOperationRun an operation a vault item offers, such as fill, authorize, confirm_transaction or add_payment_method. Some move money or use a card: confirm with the person first.Risk:destructive · Irreversible
vaults.upsertCreate a vault (a project-scoped store of secrets, cards and credentials for browsers to use), or get the existing one with that name.Risk:write