The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external Linear tickets which could potentially contain malicious instructions designed to influence the agent's code generation or shipping behavior.
Ingestion points: Linear ticket content and metadata are fetched via the /trackers tool (reference.md).
Boundary markers: The skill establishes an execution context with explicit non-goals, binary delivery checks, and active rules to limit the scope of work (doctrine.md).
Capability inventory: The skill has the ability to modify the local file system, perform git operations (branch, commit, push), and create pull requests using gh or specialized IDE tools (doctrine.md, reference.md).
Sanitization: Risk is addressed through a structured lifecycle including two mandatory code review phases (CR1 and CR2) and a requirement to print the full PR title and body for human inspection before creation.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for git operations and pull request management.
Evidence: Commands such as git rev-parse, git push -u origin HEAD, and usage of the gh CLI or Cursor's pull-request tool are specified for branch management and shipping (reference.md, doctrine.md).

just-do-it