git-bisect-forensics
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill describes the use of
git bisect run, which involves executing shell scripts or command strings (e.g.,./reproduce.shorcargo build && cargo test) to automate regression testing. This is the intended primary purpose of the git tool. - [INDIRECT_PROMPT_INJECTION]: The skill enables the agent to process data from untrusted external sources, specifically git commit history and file contents via
git logandgit blame, which creates an attack surface for instructions embedded in code or commit messages. - Ingestion points: Git commit logs, diffs, and file history (SKILL.md).
- Boundary markers: None identified in the provided instructions.
- Capability inventory: Shell command execution via
git bisect runand repository navigation. - Sanitization: No sanitization or filtering of the ingested git data is mentioned.
Audit Metadata