AI Reports

AI Reports (project navigation) shows what the agents did in this project, what it cost and who signed it off.

Tab	What you see
Overview	Spend against budget, runs, artifacts created, tasks done, estimated time saved; spend by agent and by model; spend over time; artifacts by type.
Agents	A scorecard per agent and its recent runs.
Oversight	First time right per agent (from send-backs), how long approvals waited, writes per tool.
Coverage	A heat map of document chapters × design / code / test coverage. Gaps can be turned into a task for a hired agent.
AI audit	Every item an agent created or changed — see below.

AI audit

The audit lists every AI-generated work item with:

• the agent and model, the run that created it and later runs that changed it,
• the task it was produced for,
• the review: approved by whom and when, awaiting review, or sent back.

Each run also shows the memories that shaped it (🧠 PU-…).

Below the list:

• Project memories — every lesson with who it is for, where it was learned, who proposed it, who approved or discarded it and when, and how many runs used it.
• Agent runs in sandboxes — each run in an isolated sandbox with the commands it ran, web pages it read, pushes, and its budget cap.

Use Export CSV for spreadsheets or Print / PDF for an audit file.