Activity Log

The Activity Log is your workspace's audit trail: a chronological record of who changed what, and when. It captures changes made in the dashboard by your teammates, calls made through the Data API, and actions taken by AI agents through the Churnkey MCP server — all in one place, newest first.

Use it to answer questions like "who published this cancel flow last week?", "when did our Stripe settings change?", or "what has our AI agent done in the workspace?".

Who can access it

The Activity Log is restricted to Owners and Admins (the audit_logs:read permission). Other roles do not see the page, and the same restriction is enforced on the API, so it cannot be reached by URL. See Roles and Permissions.

Finding it

Open it from the dashboard at Settings → Activity.

Reading an entry

Each row has four columns:

• When
  The timestamp of the action.
• Source
  How the change was made — one of:
  • Dashboard — a teammate acting in the web app.
  • Data API — a server-to-server Data API key. Keys are not tied to a person, so these rows show "API key (no user)" in the User column.
  • AI agent (MCP) — an AI agent acting through the MCP server. These run under a real user via per-user OAuth, so they are attributed to that person and tagged with the agent's client name.
• User
  Who performed the action — name and email — for dashboard and AI-agent (MCP) actions. Keyless Data API actions show "API key (no user)".
• Activity
  A human-readable summary, including before → after values where relevant (e.g. "stripe settings update (stackCoupons: false → true)"). For AI-agent actions, the OAuth scopes used are shown beneath the summary.

Filtering and searching

A filter row above the table lets you narrow the trail:

• Search — free-text search across the action, the acting user (name or email), and the AI-agent actor. Type a teammate's email to see everything they changed, or a keyword like stripe or recovery to find specific actions.
• Source — show only Dashboard, Data API, or AI agent (MCP) entries.
• Date range — limit to a window of time.
• Clear filters — reset everything.

The count of matching entries is shown next to the filters, and Load more pages through older results.

AI-agent attribution

When an AI agent acts through the MCP server, it authenticates as a specific Churnkey user via OAuth and inherits that user's role. Every action it takes is recorded here under that user's name, alongside the agent's client name and the OAuth scopes it used — so agent activity is just as accountable as a teammate's.

Programmatic access

The same trail is available to AI agents and scripts through the MCP get_audit_log tool, gated by the account.audit_log.read scope (Owner/Admin only). It returns the same who/what/when data — including before → after summaries and source filtering — that the Activity page displays. See the MCP server documentation.

Common questions

Why do some rows say "API key (no user)"?
Those actions were made with a Data API key, which authenticates an integration rather than a person, so there is no user to attribute. Dashboard and AI-agent (MCP) actions always show the acting user.

Can other roles see the Activity Log?
No. It is limited to Owners and Admins, both in the dashboard and on the API.

Does searching a teammate's email find their changes?
Yes. Search matches the acting user's name and email as well as the action text, so searching an email returns everything that person did.

Do AI-agent actions really show who authorized them?
Yes. MCP actions run under the authenticating user via OAuth and are logged under that user's name with the agent's client name and scopes — see AI-agent attribution.