prd-task

To convert a markdown PRD into JSON, a workflow parses the file, extracts tasks with verification steps, and writes a structured prd.json artifact to drive autonomous execution. It also moves the original PRD to a state directory for traceability.

Deterministic task decomposition is the process of parsing a markdown PRD to extract structured tasks and verification steps, outputting a machine-readable JSON artifact. This enables automated task execution, state management, and progress tracking without manual translation errors.

You can track progress by creating a dedicated state directory under .opencode/state/, where the workflow stores the original prd.md, the executable prd.json, and an empty progress.txt indicator. This setup preserves context and enables automated validation and auditing.

Your PRD must be authored in Markdown and include clear end-state focuses and verification steps. The workflow reads the markdown file, extracts these verification steps, and preserves them in the JSON output to enable automated validation.

The best way to preserve context is to move the original PRD into a dedicated state directory and generate a structured JSON artifact alongside it. This maintains traceability while keeping key files and patterns accessible for autonomous execution.

Manually translating markdown PRDs into JSON is error-prone and creates a bottleneck for automation because it requires human intervention to structure tasks and verification steps. Generating a deterministic JSON artifact eliminates this manual overhead.