The Agent Skills Directory

[COMMAND_EXECUTION]: The script 'scripts/investigate.py' executes local 'git' commands (log, blame, grep) to retrieve historical context. This is the primary intended function of the skill. Command execution uses the list-based argument format and employs the '--' separator to ensure that user-provided targets are treated as file paths rather than command-line options, mitigating potential argument injection.- [SAFE]: The skill includes a robust path validation utility ('validate_path_no_traversal') that prevents directory traversal attacks (CWE-22) by resolving paths and verifying they remain within the current working directory before any file system or git operations occur.- [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it ingests untrusted data from local files and git history.
Ingestion points: Reads content from '.agents/architecture/ADR-*.md' and 'git log' output.
Boundary markers: None explicitly used when including these snippets in the generated markdown report.
Capability inventory: Executes local 'git' via 'subprocess.run' and writes to stdout.
Sanitization: Path validation is present, though text content from files is processed as raw data. This is categorized as low risk due to the intended scope of analyzing internal project history.

chestertons-fence