The Agent Skills Directory

[SAFE]: The analysis found no evidence of malicious intent, prompt injection, or unauthorized data access. The skill's design prioritizes local repository integrity and user authorization.
[COMMAND_EXECUTION]: The skill appropriately uses Git CLI tools to perform its stated functions. All destructive operations require explicit user authorization and state recording.
[EXTERNAL_DOWNLOADS]: Documentation references the use of curl to fetch patches from GitHub. This targets a well-known service and is a standard operation for the described developer workflows.
[INDIRECT_PROMPT_INJECTION]: 1. Ingestion points: The skill reads repository history, diffs, and commit messages as outlined in SKILL.md and archaeology.md. 2. Capability inventory: The skill has the capability to run git commands and execute the local scripts/check.py file. 3. Boundary markers: The skill includes a Rules section in SKILL.md requiring the agent to inspect status and upstream before mutation. 4. Sanitization: Verification steps include running git diff --check and a structural validation script to ensure state integrity.

git-toolkit