git-commit-helper
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION] (MEDIUM): The skill instructs the agent to execute local shell commands, specifically
git diff --staged,git status, andgit commit. While these are standard git operations, direct command execution capability can be risky if the agent's reasoning is subverted. - [PROMPT_INJECTION] (HIGH): The skill is highly vulnerable to Indirect Prompt Injection (Category 8).
- Ingestion points: The skill reads untrusted external data (source code changes) via
git diff --stagedas described inSKILL.mdandreferences/git-commands.md. - Boundary markers: Absent. There are no instructions or delimiters provided to help the agent distinguish between its instructions and the content of the code diff.
- Capability inventory: The skill possesses the ability to execute subprocess commands (git) and generate content based on external input, creating a feedback loop where malicious code could influence repository state (e.g., via
git commit). - Sanitization: Absent. No validation or filtering is performed on the data returned from the filesystem before it is processed by the AI.
Recommendations
- AI detected serious security threats
Audit Metadata