git-history-recon
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses imperative language like 'treat it as the authoritative procedure' and 'executable instruction set,' commanding the AI to follow its Mermaid diagram 'exactly as written' without improvisation. This pattern is designed to override the agent's default reasoning and safety protocols to ensure strict adherence to the skill's specific workflow.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the target repository's git history, including commit messages and filenames, and formats this into a risk report for consumption by other agents. A malicious user could craft commit messages that inject instructions into downstream agents.
- Ingestion points: Git log output containing commit messages and file paths (SKILL.md Pipelines 1, 2, 6, and 7).
- Capability inventory: The skill utilizes Bash (git, mkdir, sort, uniq, wc, head, grep), Read, and Write tools.
- Boundary markers: Not present; data is inserted directly into Markdown tables without delimiters or warnings to downstream consumers.
- Sanitization: No sanitization or escaping of the mined git data is performed before inclusion in the final report.
Audit Metadata