pass-all-checks
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill parses instructions and commands from untrusted project files such as AGENTS.md, README.md, and CI configuration files. This is a vulnerability surface where a malicious repository could supply harmful commands for the agent to execute.\n
- Ingestion points: SKILL.md describes reading AGENTS.md, README.md, .github/workflows/, .gitlab-ci.yml, config/ci.rb, Gemfile, package.json, pyproject.toml, and go.mod.\n
- Boundary markers: Absent. The skill explicitly instructs the agent to trust and use commands found in these files.\n
- Capability inventory: SKILL.md instructs the agent to execute discovered shell commands (test runners, linters, gh/glab CLI tools) and perform batch edits to files. references/rails-tools.md provides a list of common CLI tools for execution.\n
- Sanitization: Absent. The skill does not validate or sanitize the commands parsed from external files before execution.\n- [COMMAND_EXECUTION]: The skill is designed to discover and execute shell commands found in the repository environment. This inherent capability allows for the execution of arbitrary commands if they are defined in the repository's configuration or documentation.
Audit Metadata