Sub 20: Harmonize existing Coordinator services to scheduling/.mjs pure-function pattern

Parent Epic

#11831 — Sub 20 of Epic. Consistency follow-up after Sub 15 (#11858 / PR #11863) established the scheduling/<task>.mjs pure-function pattern. Without this Sub, the existing 3 services remain class-instances while Subs 15/16/17 land as pure functions — Sub 18 (#11862) wire-up would need to handle two consumer shapes.

Premise

Sub 15 (PR #11863) discovered through 6 iterations that the orchestrator-internal due-trigger projections should be pure functions in scheduling/<task>.mjs, NOT class-based services. Three existing services match the wrong-shape precedent I copied initially:

• ai/daemons/orchestrator/services/SummarizationCoordinatorService.mjs — class extends Base + singleton: true + Neo.setupClass, but the class has no reactive configs, no lifecycle, no Neo class-system features beyond namespace registration. Just wraps buildSummaryTrigger pure function + getDueTask adapter
• ai/daemons/orchestrator/services/BackupCoordinatorService.mjs — same anti-pattern
• ai/daemons/orchestrator/services/PrimaryRepoSyncService.mjs — partially: getDueTask is the projection (should move to scheduling/); runTask may be genuinely service-shaped (state + execution) and stay in services/ — verify per-file

Operator's KISS framing: complexity belongs where it's earned. These services use zero Neo class-system features and should drop the machinery.

Prescription

For each of the 3:

1. SummarizationCoordinatorService.mjs → scheduling/summary.mjs:

   • Move buildSummaryTrigger (pure function — genuinely earned, has 2 sources + handover-priority branching) + getDueTask({db, state, now, summarySweepIntervalMs, getUnreadSunsetHandoversFn, markNodesAsReadFn, log}) to scheduling/summary.mjs
   • Drop class + singleton + Neo.setupClass
   • Move spec to test/.../scheduling/summary.spec.mjs
   • Delete services/SummarizationCoordinatorService.mjs

2. BackupCoordinatorService.mjs → scheduling/backup.mjs:

   • Same shape — move getDueTask to pure function
   • Move spec, delete services/ file

3. PrimaryRepoSyncService.mjs split (verify shape first):

   • If runTask has genuine state/execution responsibilities, keep services/PrimaryRepoSyncService.mjs for just runTask; move getDueTask to scheduling/primaryDevSync.mjs
   • If runTask is also a thin wrapper, fold both into scheduling/primaryDevSync.mjs and delete services/ file

Acceptance Criteria

1. ai/daemons/orchestrator/scheduling/summary.mjs + backup.mjs + primaryDevSync.mjs ship as pure-function modules
2. Each spec moves to test/.../scheduling/<task>.spec.mjs with Coordinator/Service naming dropped
3. Orchestrator's imports update: import {getDueTask as summaryGetDueTask} from './scheduling/summary.mjs' etc. (currently the orchestrator uses the singleton-method-call shape SummarizationCoordinatorService.getDueTask({...}))
4. Delete services/SummarizationCoordinatorService.mjs + services/BackupCoordinatorService.mjs (and services/PrimaryRepoSyncService.mjs if fully migrated; trim if only runTask remains)
5. All existing unit tests for the migrated services pass at the new locations + new shapes
6. No JSDoc decay-prone references (no line numbers, no section labels, no neighboring-file method names — per Sub 15 lesson)

Avoided Traps

• DON'T just rename without dropping the class machinery. The point is removing extends Base + singleton: true + Neo.setupClass, not just relocating
• DON'T preserve Neo.gatekeep registration "for tooling discoverability" — pure-function modules don't need namespace registration; consumers import directly
• DON'T fold all 3 into one file. Per-task files preserve parallel-reviewability + match the precedent Sub 15 established
• DON'T add JSDoc cross-references that decay (per Sub 15 final lesson)

Depends on

Epic #11831. Sub 15 #11858 (precedent — merged via PR #11863).

Unblocks

Sub 18 #11862 — Sub 18 can now consume ALL coordinators via the same import shape (import {getDueTask as X} from './scheduling/X.mjs') instead of needing branched logic for class-instances vs pure functions.

Authority

Operator KISS framing in #11863 merge feedback: complexity belongs where it's earned; the 3 existing services use zero Neo class-system features and should drop the machinery for consistency with Sub 15's pure-function pattern.

Authored by: [Claude Opus 4.7] (Claude Code)