Outcome
Bring the Campaign loop to the bar Formation already meets.
Problem
Four outer loops are specified. Doctrine carries 78 scenarios, a skill, and the sdd-scanner agent. Formation carries 43 scenarios, a skill, and the sdd-warden agent — it is complete, and its handoff autorun was deliberately removed in favor of an on-demand nudge, so it is under-triggered rather than under-built. Campaign and Forge have a prose README, zero scenarios, no skill, and no agent. Neither has ever run.
Campaign is the Oracle's loop and the only one that asks what the product should be. Its absence means grow and prune findings are made ad hoc in conversation, if at all.
Work
- Write the
.feature for campaign/, covering the four triggers its README already tables and the detect-and-draft versus keep-or-cut split.
- Implement the loop skill and its delegate agent, parallel in shape to
formation-loop and sdd-warden.
- Consume the forward cursor rather than cold-scanning the product.
- Record findings on the product axis rather than emitting CRs straight from a cold derivation.
- Decide explicitly whether Forge follows in this initiative or is deferred; it is the fourth loop and equally unbuilt, but it is maintainer-facing and differently shaped.
Acceptance
A campaign pass over a product with a recently shipped capability emits a grow finding naming a successor and a prune finding where one capability subsumes another, each carrying its evidence. No finding becomes work without a ratified CR, and the loop never writes a lifecycle transition itself. A second pass over unchanged state re-emits nothing already recorded resolved.
Initiative and dependencies
Project: https://github.com/orgs/cyberuni/projects/2
Parent: #16
Blocked by:
Outcome
Bring the Campaign loop to the bar Formation already meets.
Problem
Four outer loops are specified. Doctrine carries 78 scenarios, a skill, and the
sdd-scanneragent. Formation carries 43 scenarios, a skill, and thesdd-wardenagent — it is complete, and its handoff autorun was deliberately removed in favor of an on-demand nudge, so it is under-triggered rather than under-built. Campaign and Forge have a prose README, zero scenarios, no skill, and no agent. Neither has ever run.Campaign is the Oracle's loop and the only one that asks what the product should be. Its absence means grow and prune findings are made ad hoc in conversation, if at all.
Work
.featureforcampaign/, covering the four triggers its README already tables and the detect-and-draft versus keep-or-cut split.formation-loopandsdd-warden.Acceptance
A campaign pass over a product with a recently shipped capability emits a grow finding naming a successor and a prune finding where one capability subsumes another, each carrying its evidence. No finding becomes work without a ratified CR, and the loop never writes a lifecycle transition itself. A second pass over unchanged state re-emits nothing already recorded resolved.
Initiative and dependencies
Project: https://github.com/orgs/cyberuni/projects/2
Parent: #16
Blocked by: