The Agent Skills Directory

[SAFE]: The skill implements automated security scanning to identify and block hardcoded credentials such as API keys and tokens during the staging process.\n- [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for untrusted data via git logs and diffs in SKILL.md. Mitigation: The skill mandates user confirmation for all commit and push operations, preventing autonomous execution of instructions embedded in commit data. Evidence Chain: (1) Ingestion points: git fetch, log, and diff; (2) Boundary markers: user review of diff stats; (3) Capability inventory: Bash tool used for git/gh commands; (4) Sanitization: secret scanning regex.\n- [COMMAND_EXECUTION]: Uses the Bash tool to perform standard repository operations and interact with the GitHub CLI. These actions are limited to version control tasks and are governed by strict confirmation rules.

git