The Agent Skills Directory

[Command Execution] (SAFE): The skill utilizes standard git binary commands to inspect the repository and manage branches. Potentially sensitive operations, including git commit and git push, are protected by mandatory interactive user confirmation artifacts, preventing unauthorized changes or unintentional data transfer.\n- [Prompt Injection] (SAFE): No instructions designed to override agent behavior, bypass safety guardrails, or reveal system prompts were identified within the skill definition.\n- [Data Exposure] (SAFE): File access is restricted to the current Git repository via git diff. No patterns targeting sensitive system files, configuration directories (e.g., .ssh, .aws), or hardcoded secrets were detected.\n- [Indirect Prompt Injection] (SAFE): The skill is subject to a surface for indirect prompt injection via the content of staged files processed by git diff. This is inherent to the functionality of a code-summarization tool. The risk is mitigated by the mandatory human-in-the-loop review of the generated output before any commit or push is finalized.\n
Ingestion points: Staged file content accessed via git diff --cached (SKILL.md).\n
Boundary markers: Absent.\n
Capability inventory: Shell execution of the git utility for repository management.\n
Sanitization: Absent; the system relies on user verification of the generated commit message output.

commit-message