discover-execute

Automating a discovery plan requires driving a halt-loop across local knowledge-base references to synthesize citations into a structured blueprint. This process validates plan constraints and populates coverage corners before emitting the final markdown file.

A halt-loop discovery process is an autonomous research mechanism that iterates over local references, collecting evidence and validating citations on disk. It stops only when all research questions are answered or legitimately blocked, ensuring verifiable results.

Executing an approved discovery plan involves invoking the automation command with the plan slug. The system then operates on the plan's knowledge-base references, enforces boundary rules, and outputs the blueprint to the discoveries directory.

Blueprint generation validates citations on disk automatically as part of its halt-loop execution. It ensures all evidence collected from local references is verified before the final blueprint is synthesized and emitted to the file system.

When research questions are blocked during autonomous discovery, the halt-loop stops and supports human-in-the-loop oversight. It provides an auditable progress trail for iterations and citations, allowing users to review and resolve blocked questions.

An approved discovery plan is required before running the blueprint automation. The execution process relies on the plan's predefined knowledge-base references and constraints to drive the halt-loop and generate the final structured output.