The Agent Skills Directory

[SAFE]: The skill incorporates strong security defaults, including a mandatory secret scanning step that checks the branch history and proposed diff for API keys, tokens, and private keys before any publication occurs.- [SAFE]: A 'one-point approval' mechanism ensures that the user reviews the exact diff, verification results (tests/syntax), and deployment impact before the agent performs any destructive actions.- [SAFE]: The skill uses state fingerprinting (SHA-256/shasum) to verify that the repository state hasn't been modified between the user's approval and the final commit, preventing the accidental publication of unreviewed changes.- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes data from external files within the repository to generate release summaries.
Ingestion points: Reads repository files such as package.json, README.md, CLAUDE.md, and Git logs to identify project context and change history.
Boundary markers: The instructions do not specify the use of delimiters or explicit 'ignore instructions' markers when passing the contents of these files to the agent's context.
Capability inventory: The skill can execute repository-defined build scripts (e.g., npm test, npm run deploy) and perform Git operations (git commit, git push).
Sanitization: No automated sanitization is performed on the ingested file content, relying instead on the manual user approval step as a security control.- [COMMAND_EXECUTION]: The skill executes local Git commands and project-specific build tools to perform environment discovery and release integration. All such commands are performed within the local repository context and are subject to the user's final approval.

sdd-release