code-tour
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads external codebase files (README, source code, and configuration) to generate tour descriptions, which creates a surface for malicious instructions to influence agent behavior.\n
- Ingestion points: Target repository files are read during the 'Discover the repo' and 'Read every file' steps in SKILL.md.\n
- Boundary markers: Absent; the instructions do not include delimiters or warnings to ignore instructions embedded within the analyzed codebase files.\n
- Capability inventory: The skill has the capability to write files to the '.tours/' directory and execute the local 'validate_tour.py' script.\n
- Sanitization: Absent; no explicit sanitization or filtering of content read from external files is described before it is interpolated into the JSON output.\n- [COMMAND_EXECUTION]: The workflow involves running an internal Python script to validate the integrity of the generated walkthrough artifacts.\n
- Evidence: SKILL.md explicitly instructs the agent to run 'python skills/code-tour/scripts/validate_tour.py .tours/.tour --repo-root .' during the validation phase.
Audit Metadata