The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs multiple Git and GitHub CLI operations to automate the approval and branch creation workflow.\n
Evidence: Executes git checkout, git push, gh pr merge, and gh pr create as detailed in Steps 3, 4, 5, and 7 of SKILL.md.\n
External Dependency: The skill relies on an external shell script ../plot/scripts/plot-pr-state.sh. This is a vendor-provided tool that is part of the author's broader workflow.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it executes shell commands using data extracted from project files.\n
Ingestion points: Branch names and descriptions are parsed from the ## Branches section of plan files in docs/plans/active/ (Step 4).\n
Boundary markers: No explicit boundary markers or instructions to ignore embedded instructions are used when processing the plan file content.\n
Capability inventory: The extracted strings are interpolated into git checkout -b and gh pr create --title commands (Step 5).\n
Sanitization: There is no evidence of sanitization or escaping of shell meta-characters (e.g., backticks, quotes, semicolons) for the parsed strings before they are used in shell commands.

plot-approve