skills/techtechwizard/skills/commit/Gen Agent Trust Hub

commit

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute git status, git diff, git add, and git commit. This is the intended primary purpose of the skill. It correctly uses HEREDOCs when committing to preserve message formatting and prevent shell interpolation issues.
  • [INDIRECT_PROMPT_INJECTION]: The skill has a large ingestion surface as it reads data from git diff, git status, and external standards files (.claude/standards/).
  • Ingestion points: Reads project files, home directory configuration, and git output.
  • Boundary markers: The instructions tell the agent to describe what the diff does rather than what a developer says, which provides a logical boundary, though no formal delimiters are used for the diff content.
  • Capability inventory: The skill has access to Bash, Read, Grep, and Glob tools.
  • Sanitization: There is no explicit sanitization of the diff content or standards files before processing.
  • [DATA_EXPOSURE]: The skill is designed to read configuration from ~/.claude/standards/. While this involves accessing the user's home directory, it is a documented feature for retrieving team-wide or developer-specific commit conventions.
  • [PROMPT_INJECTION]: The skill contains instructions such as "This convention overrides whatever the harness suggests by default" and "NEVER add a Co-Authored-By line". These are used to enforce specific formatting standards and ensure the agent does not use platform-default templates that might conflict with the team's requirements.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 10:44 PM