The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs high-impact operations including cargo release <level> --execute, git commit, and git push. These commands modify the repository state, create tags, and push to remote origins. The risk is mitigated by a mandatory manual confirmation gate ("yes") before the final execution step.
[PROMPT_INJECTION]: The skill ingests data from local files (Cargo.toml and CHANGELOG.md) to determine versioning logic and release notes, creating an attack surface for indirect prompt injection.
Ingestion points: The agent reads the version field from Cargo.toml and the [Unreleased] section from CHANGELOG.md to compute the next release version.
Boundary markers: Absent. There are no explicit instructions to the agent to treat the file content as untrusted data or use delimiters.
Capability inventory: The skill has capabilities to commit to the repository, create git tags, and push changes to remote servers via cargo release and git CLI tools.
Sanitization: Absent. The skill recommends versions directly based on the presence of keywords (e.g., "Removed", "Fixed") within the changelog without sanitizing the input strings.

jackin-release