The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified in the skill. It performs its documented function of scanning local history files and managing a git pre-commit hook without any hidden or suspicious behavior.
[COMMAND_EXECUTION]: The skill utilizes the subprocess module in scripts/setup.py and scripts/guard.py to interact with Git and execute internal Python scripts. This execution is limited to the local environment and is necessary for managing the pre-commit hook.
[PROMPT_INJECTION]: The skill processes untrusted chat history which constitutes an indirect prompt injection surface.
Ingestion points: Local markdown files in the .specstory/history/ directory are read by scripts/scan.py.
Boundary markers: None; files are processed line-by-line using regular expressions.
Capability inventory: The skill uses Bash, Read, and Write permissions to manage the .git/hooks directory and execute local Python scripts.
Sanitization: Scan findings are reported back to the agent as truncated snippets (maximum 200 characters) to reduce the risk of processing large or malicious payloads.

specstory-guard