The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use Git, GitHub API, and vendor-specific tools like and-backend-contract to manage the lifecycle of delivery units. These operations are restricted to merging PRs and cleaning up related branches as part of the intended workflow.\n- [SAFE]: The skill reads the local configuration file .and/config.yml to resolve the backend contract and operational state, which is a standard procedure for this ecosystem.\n- [EXTERNAL_DOWNLOADS]: References to tools like and-implement, and-pack, and and-sweep are consistent with the 'Andiedie' vendor namespace and represent internal tool calls rather than downloads from untrusted sources.\n- [PROMPT_INJECTION]: The disable-model-invocation: true setting in the frontmatter provides a security boundary by preventing the model from executing tools in response to instructions found in the external review or receipt data it processes.

and-finish