checkpoint-distill
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates a workflow where potentially untrusted data from repository files is distilled into authoritative project instructions.
- Ingestion points: The skill reads Change Request documents (located in
docs/cr/), iteration ledgers (indocs/cr/iterations/), and Git commit history viagit log. - Boundary markers: The skill does not define specific technical boundary markers or use instructions to ignore embedded content when reading from the source files.
- Capability inventory: The skill has permissions to read repository files, execute Git commands (
git log,git merge-base), write to the project's standing instructions, and create new Git commits. - Sanitization: The instructions do not specify any sanitization, filtering, or escaping of the narrative text extracted from source artifacts before it is interpolated into the project's standing instructions.
- Mitigation: The risk is mitigated by a mandatory human-in-the-loop approval process, where candidates are presented in tiers and only written to disk after explicit user selection.
Audit Metadata