agent-spec-authoring

To write an AI agent task contract, you create a `.spec` file with defined intent, technical decisions, constraints, and boundaries to ensure the agent implementation is clear and machine-verifiable.

BDD-style scenario writing in agent specifications involves authoring acceptance criteria with explicit test selectors and step tables to verify agent behavior against the task contract.

You refine an existing agent-spec file by improving its intent, constraints, boundaries, and scenarios, while fixing lint warnings to enforce hard syntax rules and improve contract quality.

No specific frameworks are required to scaffold a `.spec` file. You can author task contracts independently to define API endpoints or user profile updates without external dependencies.

Your agent-spec contract fails lint warnings when it violates hard syntax rules or lacks a complete behavior surface checklist, requiring you to fix structure and rewrite parity contract considerations.