commit
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to ingest and follow instructions from external, potentially untrusted sources such as
AGENTS.md, contribution documentation, and GitHub issue metadata to determine commit conventions and task context. - Ingestion points: Files like
AGENTS.md, contribution docs, and repository metadata (commit history and issue descriptions) are read at runtime inSKILL.md(Inspect section). - Boundary markers: The instructions lack explicit boundary markers or warnings for the agent to ignore natural language instructions that might be embedded within these external files.
- Capability inventory: The skill uses
gitcommands (git status,git diff,git add,git commit) which, while standard, are controlled based on the input from the external files. - Sanitization: There is no evidence of sanitization or filtering of the content read from external repository files before the agent processes it to form its commit plan.
Audit Metadata