The Agent Skills Directory

Indirect Prompt Injection Surface: The skill evaluates several project artifacts, including specifications, plans, and task lists, to determine the status of quality gates. If these documents contain instructions intended to influence the agent's behavior, they might be processed as input data during the validation process.\n
Ingestion Points: Artifacts such as spec.md, plan.md, tasks.md, and checkpoint YAML files found in the project directory.\n
Boundary Markers: The skill utilizes structured headers and predefined checklist formats to parse these documents.\n
Capability Inventory: The skill is designed to read local files and generate validation reports; it is explicitly configured as read-only for source code files.\n
Sanitization: The build verdict logic is designed to prioritize verifiable return codes from shell commands over narrative descriptions to ensure objective assessments.\n- Network Interaction Patterns: Documentation within the skill describes verification strategies that involve interacting with API endpoints for functional equivalence testing.\n
Evidence: Reference files like references/functional-equivalence.md provide examples for comparing outputs using endpoints such as http://localhost:8080 and http://legacy:8080.

quality-gates