The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various CLI tools including git, gh, ruff, and pytest to manipulate the repository state and verify code quality.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection because it reads and processes untrusted content from Git conflict markers that originate from external branches.
Ingestion points: Conflicting files read during the resolution process in Step 5.
Boundary markers: The skill relies on standard Git conflict markers (<<<<<<<, =======, >>>>>>>) to delineate code segments.
Capability inventory: The skill can write files, perform git commit and git push, and execute arbitrary code within the repository context using pytest.
Sanitization: The skill relies on AI-based logic to resolve conflicts but lacks automated sanitization or sandboxing for the content being merged and tested.

fix-conflict