The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of local command-line tools to perform its tasks.\n
Executes git commands for repository state checks, committing changes, tagging versions, and pushing to remote origins.\n
Uses the rota CLI for specialized release operations, version calculations, and changelog management.\n
Invokes gh and glab CLI utilities to interact with GitHub and GitLab for publishing release notes and assets.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external data which could potentially contain malicious instructions intended to influence the agent.\n
Ingestion points: Fetches commit history via git and issue descriptions from tracking systems to generate automated release notes (SKILL.md, Step 5). It also reads a project-level checklist file (.rota/RELEASE.md) in Step 2.\n
Boundary markers: Includes a mandatory human review step (Step 6) where the agent presents the drafted notes for verification and approval before any destructive or remote actions (tagging, pushing, publishing) are taken.\n
Capability inventory: The skill has permissions to write to the local filesystem (CHANGELOG, version files), commit to Git, and push tags/releases to remote servers via authenticated CLI tools.\n
Sanitization: External content from commits and issues is processed directly by the model for summarization without explicit filtering or sanitization steps mentioned in the instructions.

rota-release