Skip to content
🧭
Grafton detail route

Integration Guide

Use this page when you already understand Grafton as the initiative steward and now need the shortest practical guide for creating clean Seeds, reading lifecycle-health nudges, and closing work in a way Grafton can actually harvest.
Validated Β· Oct 6, 2026The Grafton proof still matches the current Lifecycle Model, the district contract, and the phone-first exits that keep stage, coordination, and district pivots visible before the lower detail sections begin.
Use Integration Guide for intake and closeout hygieneHere

Stay here for Seed minimums, stale-work nudges, and the clean closeout path that keeps harvest honest.

Best first question β†’ what does Grafton need from this record?
Open Lifecycle Model for stage truth

Jump to the canonical Seed, Work Item, and harvest stages when the remaining question is proof or current state.

Open Lifecycle
Open Coordination for the handoff pattern

Switch routes once Grafton already clarified the lifecycle and the remaining question is how the work should move next.

Open Coordination
Pick the route

Decide whether this is really an integration question

Atlas should not make you read the wrong Grafton page. Use the quickest sorter first so intake, lifecycle semantics, and handoff mechanics stay on their owning routes.

Start on the route that answers the actual question

Grafton Integration is intentionally narrower than the old bespoke page. It covers how to work with Grafton, not every fact about Grafton.
If the question is...
I need to plant new work or clean up an existing Grafton handoff.
Start here
Stay on this page.
Why
This route explains the practical integration shape: what to hand Grafton, which gaps it routes back, and how the closeout loop behaves once execution starts.
Use this when the missing piece is not β€œwhat stage is this?” but β€œwhat must the record contain so Grafton can steward it honestly?”
If the question is...
I already have a Seed or Work Item and the real question is stage, gate proof, or stale movement.
Start here
Open Lifecycle Model.
Why
Grafton's lifecycle route owns the canonical stage semantics: Draft, Reviewed, promotion checks, Work Item health, and harvest.
Do not re-read this integration page when the actual question is β€œwhat does Reviewed mean?” or β€œwhy is this Work Item stale?”
If the question is...
Grafton already pointed to the next owner, but the handoff itself is still unclear.
Start here
Open Coordination or Capabilities.
Why
Once the lifecycle answer is clear, Atlas should move you to the route that owns owner selection or multi-agent handoff patterns instead of flattening every question into Grafton.
Integration gets the work into the lane. Coordination explains how it moves between lanes.
Seed handoff

Give Grafton a Seed it can steward honestly

The minimum creation record is intentionally light, but promotion now has explicit proof checks. Use the split below so new work lands quickly without hiding what still has to mature.
Minimums

Creation minimum versus promotion-ready minimum

At creation
Problem framing and an explicit owner are the minimum useful record. That is enough for Grafton to keep the idea visible without pretending it is already execution-ready.
Use this when the work is real but still gathering proof.
Before promotion
Add upside sizing, proof to convince, a sponsor, and the first scoped Work Item before the Seed graduates into execution.
These are current promotion checks, not optional polish.
Fast rule: Promotion is a proof check, not a momentum check.
Problem framing

At creation, give Grafton a falsifiable gap instead of a wish list. If the problem cannot be named concretely, the Seed is still shaping.

Good shape: what is wrong or missing, where it shows up, and why it matters now.
Owner

Name who owns the Seed from day one. Grafton stewards movement, but it should not have to guess whose lane the idea belongs in.

A Seed without an owner is not a lighter intake record. It is an ambiguous routing problem.
Upside sizing and proof to convince

Before promotion, add the rough leverage and the evidence that would make the move rational.

Lifecycle Model now treats these as core promotion checks rather than optional polish.
Sponsor and first scoped Work Item

A Reviewed Seed is only ready to promote when the approval owner and at least one bounded execution slice are explicit.

If the first Work Item still cannot be named, the Seed is not yet execution-ready.
Work item health

Know what Grafton treats as stale, blocked, or incomplete

The earlier page taught older timing and escalation expectations. The current route keeps the cues aligned with the canonical lifecycle and workflow docs so readers do not learn a false health signal.
Draft Seed quiet for 2+ weeks

Current Grafton guidance flags a Draft Seed back to the owner when enrichment stalls for two or more weeks.

This refresh replaces the older 5-day / 10-day escalation teaching that no longer matches the canonical lifecycle route.
Work Item with no meaningful status change for 1 week

Grafton treats one week of unmoving execution as a lifecycle-health problem, not as harmless quiet.

When the work keeps bouncing through Clarify or simply stops moving, the route should name the blocker rather than leave the item looking healthy.
Needs Guidance sits for 4+ hours

When a human decision is the blocker and the lane still has not moved after four or more hours, Grafton escalates beyond the shared awareness layer.

Atlas keeps the emphasis on the escalation threshold, not on reproducing every runtime notification mechanic.
Approved or Done without proof

Result Summary quality matters at closeout. Grafton prompts the owner when the completion proof is missing instead of pretending harvest already happened.

Done is the start of the harvest loop, not the end of responsibility.
Seed routed back for missing proof

When a Reviewed Seed lacks upside sizing, proof to convince, or the first scoped Work Item, Grafton sends it back with the named gap rather than guessing.

Execution owner notified after promotion

Once the Seed clears the gate, Grafton creates bounded Work Items and nudges the owning execution lane instead of absorbing the work itself.

Human-needed blockers stop being passive

Shared awareness comes first, then direct escalation when the blocking decision remains unresolved long enough to stall the lane materially.

Completion prompts are part of integration too

If Result Summary proof is missing, Grafton routes the closeout correction to the owner before waking Dormant Seeds or planting follow-on work.

Closeout loop

Done should trigger harvest, not just a quieter status chip

Integration also covers the handoff after execution. The work is not really closed until Grafton can verify the result, wake anything blocked on it, and decide whether the outcome should plant follow-on work.
  1. 1.Verify the Result Summary is filled. Missing proof routes back to the owner before harvest continues.
  2. 2.Check Dormant Seeds for wake conditions that this result just satisfied, and reactivate the newly unblocked ones.
  3. 3.Create linked Draft Seeds only when the result implies a real follow-on opportunity.
  4. 4.Update the parent Seed only when all related Work Items are complete so lifecycle truth stays honest.

Continue through the Grafton cluster

Grafton district

Return to the district once you want the shorter role summary instead of the integration details.

Open Grafton district
Lifecycle Model

Open the canonical stage and gate reference when the remaining question is lifecycle state rather than integration shape.

Open Lifecycle Model
Coordination

Switch here when Grafton already clarified the lifecycle and the next question is how the work should move across lanes.

Open Coordination