The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to define and execute shell commands as part of its quality assurance workflow.
Evidence: In SKILL.md, the Quality Gate Configuration section explicitly lists commands such as npm run lint, npm run typecheck, and npm test to be executed at various stages.
Evidence: In examples/milestone-m1.tasks.yaml, the validation step for task m1-006 includes running npx ts-node scripts/validate-examples.ts to verify examples.
[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection because it processes untrusted data from requirement files which directly influences the generation of tasks and commands.
Ingestion points: The skill ingests data from business-requirements.yaml, technical-requirements.yaml, and external files provided via the --style-anchors parameter.
Boundary markers: There are no explicit boundary markers or instructions defined to prevent the agent from obeying malicious instructions embedded within these requirement files.
Capability inventory: The skill has the capability to define and initiate subprocess calls (linting, testing, type checking) and write new files (milestones.yaml, milestone-m*.tasks.yaml) based on the analyzed input.
Sanitization: The skill lacks sanitization or structural validation for the content of the input YAML files before using them to establish functional scope and dependencies.

implementation-planner