The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various shell commands using git and gh (GitHub CLI) to manage tags, fetch history, and create releases. It also executes build and test commands derived from the target project's local environment.
[DYNAMIC_EXECUTION]: The skill performs dynamic command determination by reading instructions from project-specific files like CLAUDE.md or AGENTS.md. If a ## Releasing section contains a check command, the skill executes it exactly as written. It also auto-detects build scripts from package.json and composer.json and executes them using detected package managers (npm, bun, yarn, pnpm).
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the project directory which could influence its behavior if used on a malicious repository.
Ingestion points: Target repository files including CLAUDE.md, AGENTS.md, package.json, composer.json, and git tag metadata.
Boundary markers: None present; the skill trustingly parses and executes instructions found in these files as part of the release workflow.
Capability inventory: Execution of shell commands via git, gh, and various package managers. The skill can run arbitrary commands defined in the project's documentation files.
Sanitization: None present; command strings extracted from configuration files are executed without validation or escaping.

release