The Agent Skills Directory

[COMMAND_EXECUTION]: The skill extensively uses shell commands for Git operations (git fetch, git tag, git merge-tree, git ls-tree, etc.) and platform-specific CLI tools (glab, gh, tea). These commands are used to automate the entire release process, from pre-flight checks to final merging and reporting.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the Git repository environment and interpolates it into command execution.
Ingestion points: Data enters the agent's context through commands like git tag, git log, git branch, and platform API responses in SKILL.md (Phases 1, 3, 5, 8, and 9).
Boundary markers: The skill does not employ explicit boundary markers or 'ignore embedded instructions' warnings when interpolating branch names, tag names, or commit messages into shell commands.
Capability inventory: The skill has broad capabilities, including executing shell commands, writing files, and performing network operations via Git and platform CLIs across all three files.
Sanitization: While the skill uses sophisticated regex (^<{7,} |^={7,}$|^>{7,} |^\|{7,} ) to detect and block file content containing Git conflict markers (Phases 0 and 8), it lacks sanitization or validation for repository metadata (branch/tag names) used in command interpolation.
[PERSISTENCE]: The Appendix in SKILL.md provides instructions for installing a .git/hooks/pre-commit shell script. This is a persistence mechanism that ensures a specific script runs automatically during every Git commit operation within the repository.
[EXTERNAL_DOWNLOADS]: The skill recommends installing additional tooling from the author's own repository using npx skills add FuDesign2008/open-skills -g --skill '*' --yes if a dependency is missing. This represents a functional requirement for the skill to operate within the author's ecosystem.

git-release-finish