What it does
Takes a one-line idea to a complete product requirements document, and takes an existing PRD apart to show what it is missing. It writes the sections people skip — success metrics, explicit non-goals, dependencies, risks, and the open questions that are cheaper to answer now than after a sprint of building. It grounds the document in what already exists. Before writing, it searches the space for prior work on the same problem and reads the linked Jira issues, so the PRD reflects decisions already made instead of quietly contradicting them. Where it cannot ground a claim, it writes an open question rather than a confident-sounding sentence — a PRD full of unsourced certainty is the expensive failure mode. In refine mode it never overwrites anyone's document. It reads the existing PRD, finds the gaps against a fixed checklist, and leaves its critique as a comment on the page, so the author keeps ownership and the history stays intact. Every page it creates or comment it leaves is shown in full and approved before it is written.
Example prompts
- We want to improve self-serve onboarding. Check what we've already tried, then draft the PRD.
- Read the PRD on page 4915203 and tell me what's missing before we take it to engineering.
- Draft a PRD for bulk CSV import, and flag anything I haven't given you a baseline for.
Before you connect
The credentials this agent will ask you for — the full setup is on the Setup tab.
Needs 1 credential (1 required) to connect. See setup
