The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to utilize standard developer tools, specifically git fetch and git log, to synchronize with the integration branch and diagnose issues based on the latest code state.
[INDIRECT_PROMPT_INJECTION]: As the skill is designed to process and improve a project's codebase, it inherently interacts with untrusted data that could contain malicious instructions.
Ingestion points: Reads from project files including source code and documentation (e.g., PRINCIPLES.md, CONTRIBUTING.md).
Boundary markers: The skill relies on behavioral constraints and user-in-the-loop requirements rather than technical input delimiters.
Capability inventory: The agent is empowered to perform git operations and modify files to implement fixes.
Sanitization: The instructions include strong guardrails, such as mandatory user approval for product decisions, architectural changes, and database migrations, which effectively mitigate risks from untrusted code content.

fix-bug