The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs agents to process external documents such as tickets and specifications to guide code implementation. There are no defined delimiters or instructions to prevent the agent from following malicious commands hidden within these external inputs.
Ingestion points: The orchestrator reads specifications in SKILL.md (Step 1), and the implementer reads them in the Inputs section of IMPLEMENTER-BRIEF.md.
Boundary markers: No explicit markers are defined to separate instruction from external data.
Capability inventory: The agents can modify files, execute code via the TDD loop, and perform network-facing operations via the GitHub CLI.
Sanitization: No validation or sanitization of input data is performed.
[DYNAMIC_EXECUTION]: The implementer subagent is required to generate code and execute it through focused tests, typecheckers, and full test suites. This involves running dynamically created content that may be influenced by external specifications.
[COMMAND_EXECUTION]: The skill makes use of git and the GitHub CLI (gh) to manage repository state, create branches, and synchronize pull requests based on the implementation workflow.

implement