> ## Documentation Index
> Fetch the complete documentation index at: https://docs.forge.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Audit Log

> Search organization security events and verify their tamper-evident integrity.

The Audit Log is the organization-scoped record of security and administrative
activity in Forge. It identifies the action, actor, target, request context, and
state change behind each event and presents the newest events first.

## Event schema

Every audit event contains an ID, organization, action, actor type, occurrence
time, creation time, and JSON objects for before values, after values, and
metadata. Events can also include:

| Field         | Meaning                                                                       |
| ------------- | ----------------------------------------------------------------------------- |
| Actor         | Console user ID, role, and display name when the action has a human actor.    |
| Target        | Type, stable ID, and display name of the affected resource.                   |
| Request       | Request ID, source IP address, and user agent.                                |
| Before values | Security-relevant state before a mutation.                                    |
| After values  | Security-relevant state after a mutation.                                     |
| Metadata      | Action-specific identifiers, status, revision, or bounded execution evidence. |

The Console table shows the occurrence time, action, actor, target, and request
ID or IP address. Open any event to inspect its available details. The audit
API returns the complete typed event, including before values, after values,
metadata, and user agent.

### Event details

Open an event to review its actor, target, request context, and approved state
changes. The JSON view can be copied and contains exactly the sanitized fields
shown in the drawer. Action-specific values that are not approved for Console
display remain omitted.

Policy audit snapshots contain revision, state, ownership, management mode, and
definition hash information. Use policy revision history for complete policy
content.

### Actor types

| Type             | Source                                               |
| ---------------- | ---------------------------------------------------- |
| `console_user`   | An authenticated Forge Console user.                 |
| `system`         | A Forge-owned system operation.                      |
| `workos_webhook` | An identity or directory event received from WorkOS. |
| `background_job` | A scheduled or asynchronous Forge operation.         |

## Covered activity

Audit actions are namespaced strings such as `member.*`, `auth.*`, `sso.*`,
`retention.*`, and `mcp_registry.*`. Recorded activity includes:

* Membership, role, invitation, SSO, and credential changes.
* Policy, approval, governance, and remediation decisions.
* MCP Registry servers, installations, OAuth grants, and gateway tokens.
* Integration, export destination, and delivery changes.
* Retention settings, preservation holds, and cleanup outcomes.
* Configuration publication, rollback, and endpoint operations.

An audit event records a completed or attempted Forge operation that explicitly
emits audit evidence. It is not a replacement for the higher-volume session and
gateway telemetry shown in [Sessions](/observe/sessions).

## Search

The Console supports:

* After and before timestamps.
* Action categories.
* Actor type.
* Target type.
* Newest-first pagination with 25 events per page.

The API additionally supports an exact console actor ID, exact target ID,
offset or cursor pagination, and up to 200 events per request. An `action`
filter is exact unless it ends in `.`, in which case it matches that namespace
prefix.

## Integrity

Protected organization audit events, governance evaluations, and AI
control-point events are appended to an organization-scoped SHA-256 hash
ledger.

Each ledger entry records:

* A monotonic sequence number.
* Source kind and source ID.
* Canonical schema version and payload hash.
* Previous-entry hash and current entry hash.

For audited mutations, the product change and organization audit event are
committed in the same database transaction. The audit event's ledger entry is
appended in that transaction as well. Ordinary updates and deletions of ledger
rows are rejected by the database.

A scheduled verifier checks sequence continuity, previous hashes, entry hashes,
canonical source hashes, and missing protected source rows. Verification
failures create a system audit event on the transition into a failed state.

### Proof packages

Forge can produce a bounded integrity proof package containing:

* Current verification status and failures.
* Ledger entries and canonical versions.
* Checkpoints spanning sequence ranges.
* Checkpoint signatures and external anchor receipts when configured.

The proof establishes the integrity of the canonical fields protected by each
schema version. It does not claim that expired prompt, response, tool, or source
detail remains available after its retention window. Newer canonical versions
deliberately protect retention-compatible facts while preserving earlier ledger
entries.

## Retention

Audit events belong to the security-record retention class. The selected
organization window is 1, 3, or 7 years, subject to the effective minimum shown
in the Console. Audit-ledger proofs preserve the integrity history defined by
their canonical versions even when separately retained evidence detail expires.

See [Privacy](/secure/privacy) for retention jobs, tombstone proofs, and
preservation holds.

## Exports

Configured Splunk, Amazon S3, and other security-export destinations can receive
selected organization audit events. The Audit Log links to active destinations
so their schemas, credentials, health, and delivery state can be managed
separately.

## Related pages

<CardGroup cols={2}>
  <Card title="Access" icon="key-round" href="/secure/roles">
    Review the identities and roles represented in audit events.
  </Card>

  <Card title="Privacy" icon="lock" href="/secure/privacy">
    Configure security-record retention and preservation holds.
  </Card>

  <Card title="Compliance" icon="file-check" href="/secure/compliance">
    Use audit evidence in framework assessments.
  </Card>

  <Card title="Splunk" icon="share-2" href="/integrations/splunk">
    Export selected security events to Splunk.
  </Card>
</CardGroup>
