ci-fix
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill has a potential surface for indirect prompt injection because it analyzes external data from CI logs to implement code fixes. \n
- Ingestion points: External data enters the context through
gh run view --log-failedand other log extraction commands inSKILL.md. \n - Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for the log data processing. \n
- Capability inventory: The skill utilizes
git pushto modify the repository andgh run rerunto trigger actions, as shown inSKILL.md. \n - Sanitization: There is no mention of sanitizing or escaping the content retrieved from the logs. \n
- Mitigation: The presence of "Safety Notes" regarding minimal permissions and secure workflow practices reduces the risk of accidental exploitation.
Audit Metadata