Agent-use audit report template
Agent-use audit: <target>
Overall: <score>/100 (<grade>) Date: <date> Reviewer: <name/agent> Surfaces reviewed: <repo/docs/CLI/API/SDK/MCP/UI/files/skill> Surfaces not reviewed: <limits>
Executive summary
<3-7 sentences. State the main strengths, biggest blockers, and highest-leverage fixes.>
Applicability
| Surface/dimension | Applies? | Why |
|---|---|---|
| Web/docs | yes/no | <reason> |
| CLI/TUI | yes/no | <reason> |
| API/SDK | yes/no | <reason> |
| App UI | yes/no | <reason> |
| Tools/MCP/A2A | yes/no | <reason> |
| Files/workspace | yes/no | <reason> |
Scores
| Dimension | Applicable | Score | Evidence | Main gap |
|---|---|---|---|---|
| Discoverability | yes/no | 0-10 | <file/url/command> | <gap> |
| Content readability | yes/no | 0-10 | <file/url/command> | <gap> |
| Capability contracts | yes/no | 0-10 | <file/url/command> | <gap> |
| Action parity | yes/no | 0-10 | <file/url/command> | <gap> |
| Context parity | yes/no | 0-10 | <file/url/command> | <gap> |
| Parseable outputs | yes/no | 0-10 | <file/url/command> | <gap> |
| Safety/recovery | yes/no | 0-10 | <file/url/command> | <gap> |
| Evals/maintenance | yes/no | 0-10 | <file/url/command> | <gap> |
Action/context parity map
| Human capability | Human context | Agent-readable context | Agent action path | Safety level | Recovery path | Gap |
|---|---|---|---|---|---|---|
| <workflow> | <UI state> | <endpoint/file/tool> | <command/API/tool> | <level> | <undo/status> | <gap> |
Findings
Critical
- <Issue> —
<evidence>— Impact: <impact>. Fix: <patchable recommendation>. Acceptance test: <test>.
High / Medium / Low
<Repeat as needed.>Top remediation plan
0-2 days
- <Fix, owner, acceptance test>
2-7 days
- <Fix, owner, acceptance test>
7-30 days
- <Fix, owner, acceptance test>
Evals to add
| Task | Current failure mode | Passing condition |
|---|---|---|
| <task> | <failure> | <assertion> |