github-repo-health-weapon

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill processes untrusted data from the repository being audited, which could be manipulated by an attacker to influence the agent's findings or behavior.
  • Ingestion points: Untrusted content is ingested from the target repository via git log for commit messages (guides/03-commit-quality.md), file system reads for README/docs (guides/06-docs-presence.md), and inspection of CI workflow files (guides/05-ci-workflows.md).
  • Boundary markers: The skill does not define clear delimiters or boundary markers to separate untrusted repository content from the agent's internal instructions.
  • Capability inventory: The skill utilizes shell commands (gh, git, ls, grep, awk) and has the capability to generate and save audit reports to the local file system.
  • Sanitization: No explicit sanitization, escaping, or filtering of repository-provided text is performed before it is analyzed by the agent.
  • [COMMAND_EXECUTION]: Shell Command Surface. The skill relies extensively on shell commands (gh api, git log, ls, grep) to gather metadata. While these are intended for read-only auditing, the interpolation of variables like {owner}, {repo}, and {branch} into command strings represents a potential command injection surface if the agent does not properly sanitize these inputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 03:22 AM
Security Audit — agent-trust-hub — github-repo-health-weapon