The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts packaged as zipapps (melt.pyz) and standard development tools such as git, mergiraf, and kdiff3 to analyze and resolve merge conflicts.
[INDIRECT_PROMPT_INJECTION]: The skill reads and processes conflict hunks from source code files, which constitutes an ingestion point for untrusted data.
Ingestion points: conflict-summary and batch-resolve commands read file content containing Git conflict markers (SKILL.md).
Boundary markers: No explicit boundary markers or instructions to ignore embedded instructions within conflict data are specified in the protocol.
Capability inventory: The skill has the capability to write to the filesystem (cp), stage files (git add), and continue Git operations (git merge --continue), alongside executing local scripts (SKILL.md).
Sanitization: The instructions do not describe any sanitization or filtering of the content within conflict markers before it is presented to or processed by the agent.

melt