The Agent Skills Directory

[SAFE]: The skill operates entirely within the local git repository context and follows secure practices, such as pinning the collation order and using stage numbers to prevent conflict resolution errors. No network operations or unauthorized system accesses were detected.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from git conflict markers through git cat-file blob calls in scripts/classify-conflicts.sh and scripts/resolve-union.sh. While this creates an indirect prompt injection surface, it is a functional requirement for conflict resolution. The scripts provide clear boundary markers in their output, and the agent's capabilities are limited to local repository modifications. No sanitization is performed on the raw diff content, but mechanical resolutions are validated.\n- [DYNAMIC_EXECUTION]: The scripts/resolve-union.sh script generates and executes a static Python heredoc to verify that the union of two file versions is sorted and lossless. This represents a low-severity dynamic execution pattern used for integrity checking rather than processing external or untrusted code.

codex-resolve-rebase-conflicts