The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill manages documentation by parsing user-provided markdown files and YAML frontmatter. This creates an attack surface where malicious instructions embedded in document metadata could influence agent behavior when scripts or summaries are processed. Ingestion points: scripts/index_docs.py, scripts/archive_docs.py, and scripts/validate_doc_metadata.py. Boundary markers: Absent. Capability inventory: Local file operations (read, write, delete) and subprocess execution via agent instructions. Sanitization: All scripts use yaml.safe_load() for secure metadata parsing, mitigating unsafe deserialization risks.
[Safe Operations] (SAFE): No hardcoded credentials, external network requests, or obfuscation techniques were detected. The skill's activities are constrained to the local documentation directory.

cyberarian