Audit actions
Every action the audit log records, grouped by the resource it acts on.
Each audit event carries an action naming what happened. The values are
stable identifiers: new ones are added as the platform grows, so treat the
list as open and ignore an action you do not recognise rather than failing.
The console shows the plain-language meaning below; this page is the reference for the machine-readable value behind it.
Agents
Creating, changing, and sharing agents.
| Value | Meaning |
|---|---|
agent.create | Someone created an agent. |
agent.delete | Someone deleted an agent. |
agent.duplicate | Someone copied an agent into a new private one. |
agent.update | Someone changed an agent's configuration. |
agent.share | Someone gave other people access to an agent. |
agent.unshare | Someone withdrew other people's access to an agent. |
agent.unattended_grant_change | Someone changed which tools an agent may use without asking first. |
agent.tool_approval_change | Someone changed which of an agent's tools need approval. |
Teams
Groups of agents that work together.
| Value | Meaning |
|---|---|
team.create | Someone created a team of agents. |
team.delete | Someone deleted a team. |
team.update | Someone changed a team's configuration. |
team.member_add | An agent joined a team. |
team.member_remove | An agent left a team. |
Pipelines
Multi-step flows built from agents and teams.
| Value | Meaning |
|---|---|
pipeline.create | Someone created a pipeline. |
pipeline.delete | Someone deleted a pipeline. |
pipeline.update | Someone changed a pipeline's steps or configuration. |
Runs
The lifecycle of work, and the decisions people made about it.
| Value | Meaning |
|---|---|
run.admitted | A run passed its limit and policy checks and was queued. |
run.start | A run began working. |
run.complete | A run finished its work. |
run.fail | A run stopped before finishing its work. |
run.cancel | Someone stopped a run before it finished. |
run.delete | Someone deleted a run's record. |
run.blocked | A policy or limit stopped a run from proceeding. |
run.resume | A paused run continued. |
run.replan | A run was sent back to produce a new plan. |
run.plan_approve | Someone approved a run's plan so work could continue. |
run.plan_deny | Someone rejected a run's plan. |
run.takeover_attach | Someone took manual control of a run's sandbox. |
run.takeover_release | Someone handed manual control of a sandbox back to the run. |
run.tool_approval_mode_change | Someone changed how a run asks for tool approvals. |
Sandboxes
The isolated environments runs work inside.
| Value | Meaning |
|---|---|
sandbox.create | A run got an isolated environment to work in. |
sandbox.create_failed | A run could not get an isolated environment to work in. |
sandbox.destroy | A run's isolated environment was torn down. |
sandbox.pause | A run's environment was suspended, keeping its state. |
sandbox.resume | A suspended environment was brought back. |
Artifacts
Files runs produce, share, and consume.
| Value | Meaning |
|---|---|
artifact.create | A run produced an artifact. |
artifact.bind | An artifact was made available to a run. |
artifact.delete | Someone deleted an artifact. |
artifact.delete_blocked_in_use | A deletion was refused because a run was still using it. |
artifact.materialize | An artifact's contents were placed in a run's sandbox. |
artifact.promote | An artifact was promoted to a durable, shareable copy. |
artifact.read_text | An artifact's text contents were read. |
artifact.binding_release | A run gave up its access to an artifact. |
Knowledge
The files agents can consult.
| Value | Meaning |
|---|---|
knowledge.file_upload | Someone added a file to the workspace's knowledge. |
knowledge.file_delete | Someone removed a file from the workspace's knowledge. |
knowledge.folder_create | Someone created a folder to organise knowledge files. |
knowledge.folder_delete | Someone deleted a knowledge folder. |
Tools and integrations
What agents can call, and the values they call it with.
| Value | Meaning |
|---|---|
user_tool.create | Someone added a custom tool agents can use. |
user_tool.delete | Someone deleted a custom tool. |
user_tool.update | Someone changed a custom tool's configuration. |
user_tool.auth_rotate | Someone replaced the credentials a custom tool uses. |
user_tool.auto_disable | The platform disabled a custom tool after repeated failures. |
mcp_server.created | Someone connected an external MCP tool server. |
mcp_server.updated | Someone changed an MCP server's configuration. |
mcp_server.deleted | Someone disconnected an MCP tool server. |
mcp_server.published | An MCP server was made available to the workspace. |
mcp_server.unpublished | An MCP server was withdrawn from the workspace. |
env_var.create | Someone added a value agents can read at runtime. |
env_var.update | Someone changed a value agents read at runtime. |
env_var.delete | Someone removed a value agents read at runtime. |
Automation
What sets work off without a person asking.
| Value | Meaning |
|---|---|
schedule.create | Someone set work to run on a schedule. |
schedule.delete | Someone deleted a schedule. |
schedule.update | Someone changed when or what a schedule runs. |
schedule.trigger | A schedule reached its time and started work. |
schedule.auto_disable | The platform disabled a schedule after repeated failures. |
inbound_webhook.create | Someone created an endpoint external systems can call. |
inbound_webhook.delete | Someone deleted an inbound webhook endpoint. |
inbound_webhook.update | Someone changed an inbound webhook's configuration. |
inbound_webhook.token_rotate | Someone replaced an inbound webhook's token, invalidating the old one. |
inbound_webhook.trigger | An external system called an inbound webhook and started work. |
inbound_webhook.auto_disable | The platform disabled an inbound webhook after repeated failures. |
agent_email.send | An agent sent an email. |
iac.channel_create | Someone opened a channel for agents to communicate on. |
iac.channel_delete | Someone deleted an agent communication channel. |
iac.message_publish | A message was posted to an agent communication channel. |
iac.signal_send | A signal was sent directly to an agent. |
Workspace and membership
The workspace itself and who belongs to it.
| Value | Meaning |
|---|---|
org.create | Someone created a workspace. |
org.delete | Someone deleted a workspace. |
org.update | Someone changed a workspace's settings. |
org.member_invite | Someone invited a person to the workspace. |
org.member_invite_revoke | Someone withdrew an invitation before it was accepted. |
org.member_accept | Someone joined the workspace they were invited to. |
org.member_remove | Someone removed a person from the workspace. |
org.member_role_change | Someone changed what a member is allowed to do. |
org.member_storage_limit_update | Someone changed how much a member may store. |
Access and credentials
Who and what can act, and how that access is granted.
| Value | Meaning |
|---|---|
api_key.create | Someone created a key for programmatic access. |
api_key.revoke | Someone revoked an API key, ending its access immediately. |
service_principal.create | Someone created a non-human identity for automation. |
service_principal.delete | Someone deleted a service account. |
service_principal.update | Someone changed a service account's configuration. |
service_principal.revoke | Someone ended a service account's access immediately. |
delegated_token.exchange | An application obtained a token to act on someone's behalf. |
delegated_token.revoke | A token acting on someone's behalf was cancelled. |
oauth_client.create | Someone registered an application that can request access. |
oauth_client.delete | Someone deleted a registered application. |
oauth_client.update | Someone changed a registered application's configuration. |
oauth_client.revoke | Someone ended a registered application's access. |
oauth_client.grant_create | Someone let an application act on their behalf. |
oauth_client.grant_revoke | Someone stopped letting an application act on their behalf. |
Governance
The rules that constrain what agents may do.
| Value | Meaning |
|---|---|
model_policy.created | Someone restricted which models may be used. |
model_policy.updated | Someone changed which models may be used. |
model_policy.archived | Someone retired a model policy without deleting its history. |
model_policy.purged | An archived model policy was permanently removed. |
model_policy.agent_create_rejected | An agent could not be created because its model is not allowed. |
model_policy.agent_update_rejected | An agent could not be changed because its model is not allowed. |
egress_policy.created | Someone restricted which sites agents may reach. |
egress_policy.updated | Someone changed which sites agents may reach. |
egress_policy.archived | Someone retired a network policy without deleting its history. |
egress_policy.purged | An archived network policy was permanently removed. |
resource.limit_exceeded | Work was refused because a usage limit had been reached. |
Budgets
Spending ceilings and the changes made to them.
| Value | Meaning |
|---|---|
budget.create | Someone created a spending budget. |
budget.update | Someone changed a budget's thresholds. |
budget.delete | Someone deleted a budget. |
Billing
Plans, subscriptions, and payments.
| Value | Meaning |
|---|---|
billing.customer_create | A billing account was opened for the workspace. |
billing.subscription_create | The workspace started a paid plan. |
billing.subscription_update | The workspace's paid plan changed. |
billing.subscription_cancel | The workspace's paid plan was cancelled. |
billing.invoice_paid | A payment for the workspace succeeded. |
billing.invoice_failed | A payment for the workspace did not go through. |
Platform administration
Actions taken by platform operators rather than inside a workspace.
| Value | Meaning |
|---|---|
platform_user.create | A platform operator created a user account. |
platform_user.update | A platform operator changed a user's details. |
platform_user.activate | A platform operator re-enabled a user account. |
platform_user.deactivate | A platform operator disabled a user account. |
platform_user.lock | A platform operator locked a user out of their account. |
platform_user.unlock | A platform operator restored a locked user's access. |
platform_user.verify | A platform operator confirmed a user's email address. |
platform_user.soft_delete | A platform operator deleted a user, keeping their records. |
platform_user.restore | A platform operator undid a user deletion. |
platform_user.hard_delete | A platform operator erased a user and everything belonging to them. |
platform_user.oauth_provider_remove | A platform operator unlinked a user's sign-in provider. |
durable_job.requeue | A platform operator retried a stalled background job. |
webhook_delivery.retry | A failed webhook delivery was attempted again. |
model_status.override_set | A platform operator changed whether a model is available fleet-wide. |
model_status.override_clear | A platform operator removed a fleet-wide model override. |
runtime_profile_deny.create | A platform operator blocked a tool from running. |
runtime_profile_deny.revoke | A platform operator lifted a block on a tool. |
signup_approval.grant | A platform operator admitted someone to the platform. |
signup_approval.deny | A platform operator refused someone access to the platform. |
signup_approval.delete | A platform operator withdrew an earlier sign-up decision. |