skill-health
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from logs and state files which could contain untrusted input from other skills that have failed or produced unexpected output.
- Ingestion points: The skill reads from
memory/logs/YYYY-MM-DD.md,memory/cron-state.json, andmemory/skill-health/*.jsonto gather failure signatures and status updates (SKILL.md). - Boundary markers: No specific delimiters or boundary markers are used when parsing log content or error signatures; the skill uses regex patterns to extract exit taxonomy markers (SKILL.md).
- Capability inventory: The skill possesses file-writing capabilities to the
memory/issues/andoutput/articles/directories, and executes shell commands includinggh(GitHub CLI) for PR management,./notifyfor external alerts, and./scripts/skill-runsfor data collection (SKILL.md). - Sanitization: There is no evidence of sanitization, escaping, or validation of the
last_errorsignatures or log snippets before they are interpolated into YAML frontmatter for new issues or the body of notifications.
Audit Metadata