Activity log & audit trail

Every project keeps a complete, automatic activity log - an append-only record of everything that happened, newest first. You never switch it on, and nothing can quietly edit or delete it after the fact. When you need to know who did what, and when - a person or an agent - this is where you look.

What gets recorded

The log captures every state-changing action on the project:

• Tasks - created, renamed, status changes, reassignments.
• Agent runs - each run started and completed, with its outcome.
• Documents - created, updated, and deleted, including edits to an agent's instructions.
• Assets - files uploaded to the project or a task.
• Agents - hired, updated, enabled, and disabled.
• Connectors - MCP connections and OAuth account connections changing.

Each entry names the actor (you, an agent, the system, or an external MCP client), so a human action and an automated one are always told apart.

Reading the log

Open a project and choose Activity in the sidebar. Entries are listed newest-first with three columns:

• Time - when it happened.
• Actor - who did it. A badge flags human admins and external MCP clients (API keys); the actor's name is always shown.
• Activity - a plain-language description ("Created task TO-4", "Changed status of TO-4 from Backlog to In Progress"). Most rows link straight to the task, agent, or page they concern.

You can narrow the view by entity type, action, and date range.

The log loads the most recent entries first, in pages. Load older activity at the bottom of the list fetches the next page and appends it, so you can walk back through the whole history as far as you need without waiting for it all up front.

Project and instance views

There are two scopes:

• Project Activity - everything for a single project, on its Activity page. This is what your project team uses day to day.
• Instance Activity - an Admin-only view under instance settings that combines the activity of all projects plus instance-level admin actions that aren't tied to any one project (managing credentials, connectors, and skills). It's the single place to reconstruct what happened across the whole instance, and adds a Project column so you can see where each action belongs.

Why it matters

The activity log makes Hezo's autonomy accountable. Agents act on their own, but every state-changing action they take is on the record, attributed, and impossible to rewrite. When something looks off, you can trace exactly what happened and roll back from a position of knowledge.