The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard command-line interfaces such as git, gh, and glab for repository management and pull request operations.\n- [COMMAND_EXECUTION]: An external utility named no-tropes is invoked to process generated text before submission. This tool is not a standard system utility.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted pull request descriptions retrieved from external servers.\n
Ingestion points: Existing PR/MR bodies retrieved via gh pr view and glab mr view in SKILL.md.\n
Boundary markers: None present to isolate untrusted PR content from instructions.\n
Capability inventory: Full git operations (reset, commit, force-push) and PR lifecycle management via gh/glab.\n
Sanitization: No explicit programmatic sanitization or delimiters are defined for the ingested content.

pr-update