The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from source code files and Git diff outputs, which creates a potential surface for indirect prompt injection if malicious instructions are embedded within the code being merged.
Ingestion points: Git commands such as git diff and git show are used to read file contents containing conflict markers.
Boundary markers: The instructions do not define specific delimiters to separate code content from instructions to the agent.
Capability inventory: The skill uses standard local tools including git and npm test to modify the repository state and verify changes.
Sanitization: There is no evidence of automated sanitization for the code content being processed.
[COMMAND_EXECUTION]: The skill involves the execution of standard Git and development environment commands.
Evidence: Commands like git status, git add, git commit, and npm test are used throughout the resolution workflow. These are appropriate and expected for a version control utility and do not represent unauthorized or dangerous system modifications.

merge-conflict-surgeon