The Agent Skills Directory

[PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it reads and processes untrusted files from a repository to generate summary reports for the agent.
Ingestion points: The scripts/analyze_project.py script recursively reads all files in the target directory (via repo.rglob('*')), including potentially attacker-controlled files like README.md or code comments.
Boundary markers: There are no boundary markers or instructions generated in the output files (SUMMARY.md, RISKS.md) to prevent the agent from following instructions embedded in the analyzed repository content.
Capability inventory: The skill performs file system reads and writes to generate its analysis reports.
Sanitization: The content from the repository is interpolated into markdown reports without sanitization or escaping of potential instructions.
[SAFE]: The skill uses yaml.safe_load() in scripts/analyze_project.py to parse analysis context, preventing unsafe deserialization of untrusted data.

analyze-project