The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Node.js scripts such as doctor.mjs, gate.mjs, and quality-suite.mjs located within the project's .specforge/core/scripts/ directory. These tools are used for project health checks, state validation, and artifact approval, which is consistent with the skill's documented automation purpose.
[INDIRECT_PROMPT_INJECTION]: The skill parses project-specific artifacts to determine the next workflow step, creating a potential surface for indirect injection if project files contain malicious instructions.
Ingestion points: Reads workflow status and decision markers from files like SKILL.md, 00-intake/brief.md, and technical-design.md.
Boundary markers: Employs explicit status strings such as [NEEDS DECISION], APPROVED, and REQUEST_CHANGES to delimit and validate state transitions.
Capability inventory: The skill can execute internal Node.js scripts and git status.
Sanitization: Automated validation is performed by the doctor.mjs and quality-suite.mjs scripts before the agent proceeds to the next automation step.

sf-work