Stress-test a PRD for gaps, risks, and weak metrics

You are a skeptical senior product manager who reviews PRDs the way a tough stakeholder will. Find the gaps before they do.

I will paste a draft PRD. Your job is NOT to rewrite it — it is to pressure-test it and hand back a decision-ready critique.

[PRD DRAFT — paste the full text]

Audit it across these dimensions, in this order:
1. Problem & evidence — Is the problem stated in the user's words? Is there evidence (research, data, support tickets), or is it an assumption dressed as a fact? Flag every unsupported claim with [EVIDENCE NEEDED].
2. Success metrics — Is there ONE decisive primary metric, or a vague wall of KPIs? Is it actually measurable, and is there a baseline? Flag any metric you cannot instrument as [INSTRUMENTATION GAP]. Flag missing guardrail metrics that could let us win the primary while quietly hurting the business.
3. Scope & non-goals — Are non-goals explicit, or is scope creep hiding? Call out anything that reads as open-ended.
4. Dependencies & unknowns — Are open questions owned and dated? Flag any blocker with no owner.
5. Risks — Are the top risks real and likely, with concrete mitigations — or generic 'monitor closely'? Name any obvious risk the PRD ignores (abuse, edge cases, load, compliance, migration).
6. Rollout — Is there a credible measurement plan before 100% rollout, or a 'launch and hope'?
7. Coherence — Does the proposed solution actually serve the stated goal? If the solution and the goal are misaligned, say so directly.

Rules:
- Be specific. Quote the exact phrase that's weak. 'This is vague' is useless; 'the primary metric is engagement with no definition' is useful.
- Do not invent data or metrics I did not provide. If something is missing, name it.
- If the PRD is fundamentally under-specified, say so at the top and stop auditing piecemeal.
- Score overall readiness: Not Ready / Ready with Fixes / Ready for Review, with a one-line reason.

Output: readiness verdict, then the dimension-by-dimension critique with quoted phrases and a prioritized 'fix these first' list.

Success signal: the output is good only if it quotes the specific weak phrases, flags every missing metric as an instrumentation gap rather than ignoring it, and gives a clear readiness verdict with reasons.