Skip to main content

Definitions of Ready / Done

A document type has two quality bars: one before anyone starts writing the body (DoR), one before anyone acts on it (DoD). Both are enforced by review, not script — the kit just gives each doc type a place to put the checklist.

Why this, not a policy wiki

Wiki policies die in a drawer. Per-type checklists live next to the doc, show up in every review, and give reviewers a concrete question to ask besides "does this look OK?".

The standard bars

Requirement card (REQ-*)

Ready to intake — the reporter has provided enough context:

  • Background / pain point / affected role are all present
  • One-liner problem statement is unambiguous
  • Rough priority signal (Must / Should / Nice)

Done (card is shipped to PM):

  • Impact analysis aligned against the Ontology / domain model
  • Complexity estimated (S / M / L / XL)
  • Assigned PM owner
  • doc_id issued

PRD (PRD-*)

Ready to write:

  • Source requirement card is Approved
  • Ontology lookup complete (upstream systems, affected entities identified)
  • No conflicting PRD exists (search the traceability matrix + Confluence)
  • Stakeholder list + decision-maker identified

Done (ready to publish):

  • doc_id, related.requirement, related.module all filled
  • Every Must / Should / Could / Won't item has testable acceptance criteria
  • Every success metric is SMART (specific, measurable, achievable, relevant, time-bound)
  • Assumptions and risks listed with mitigations
  • Three reviewers signed off (typical: PM Lead, Architect, lead engineer)
  • Status moved from DraftIn Review or Approved

Spec (SPEC-*)

Ready to write:

  • Corresponding PRD is Approved
  • Technical owner confirmed
  • Upstream API / module dependencies inventoried
  • Relevant ADRs surfaced; new-decision ADRs drafted if needed

Done:

  • doc_id, related.prd, related.module filled
  • API contract complete with error codes and examples
  • Data model aligned with Ontology
  • Test strategy covers unit / integration / performance
  • Deploy + rollback plans are executable, not aspirational
  • Architect + one senior engineer signed off

Plan (PLAN-*)

Ready to write: whoever's on point has enough context and authority to commit a team.

Done:

  • Milestones have dates or explicit "TBD pending X"
  • Dependencies are named (both what it needs and what it unblocks)
  • Rollback or abort condition is written down

How to adapt the bars

You will have types this doesn't cover (decision docs, RFCs, ops runbooks). Add a checklist at the bottom of the relevant skill / template and link the skill or template from this page.

Don't copy every checkbox into every team policy. Pick the four that fail most often in your team's review history and enforce those.

How the kit uses these

The PM skills shipped with Claude Code (see Templates: Claude Skills — full skill catalog is on the roadmap) include the relevant checklists inline. A reviewer running /create-prd ends up with the DoR/DoD baked into the generated document.

If you're not using Claude Code, paste the checklist from this page into your PR template or the doc body itself — the goal is "reviewer has to consciously sign it off", not automation.