Metadata
| Status | done |
|---|---|
| Assigned | agent-957 |
| Agent identity | eea940a6f6be13d60578dee27be1f4bade4fcaab05bbbe54b9c5ef4b2d05eae0 |
| Created | 2026-04-28T21:54:51.413162671+00:00 |
| Started | 2026-04-28T21:56:23.805479214+00:00 |
| Completed | 2026-04-28T21:59:08.386248514+00:00 |
| Tags | eval-scheduled |
| Eval score | 0.93 |
| └ blocking impact | 0.95 |
| └ completeness | 0.95 |
| └ coordination overhead | 0.93 |
| └ correctness | 0.95 |
| └ downstream usability | 0.92 |
| └ efficiency | 0.90 |
| └ intent fidelity | 0.82 |
| └ style adherence | 0.88 |
Description
Quality Pass: Post-Triage Review
Review and optimize task metadata for newly created tasks before they enter execution.
Tasks to review
- diagnose-tui-new
- fix-tui-new
What to do
For EACH task listed above:
1. Classify task type
Read the task via wg show <task-id>. Classify as one of:
- research — Investigation, analysis, library evaluation
- implementation — New code, features, endpoints
- fix — Bug fixes, error corrections
- design — Architecture, API design, planning
- test — Test writing, test infrastructure
- docs — Documentation, comments, guides
- refactor — Code restructuring without behavior change
(Hint: diagnose-tui-new is research; fix-tui-new is fix.)
2. Assign agent identity
Run wg agency stats --by-task-type to see role performance by task type.
Use the recommended role for the task's classified type. If '(insufficient data)',
fall back to the overall Role Leaderboard.
For JSON: wg agency stats --by-task-type --json →
.task_type_breakdown.recommendations[].best_role.
Apply: wg assign <task-id> <agent-hash>
3. Select model tier
Check the Best Model by Task Type table from the same stats command. Override heuristics:
| Signal | Model |
|---|---|
| Simple, mechanical, well-defined | haiku |
| Standard implementation, testing, research | sonnet |
| Complex design, multi-system reasoning | opus |
| Task has failed before | escalate one tier |
Apply: wg edit <task-id> --model <tier>
4. Release for execution
After assigning agent and model: wg resume <task-id>
Validation
-
Both tasks have an agent assigned (check via
wg show) - Both tasks have a model set
- Both tasks are un-paused (status: open, not paused)
- Assignments are justified by evaluation data, not arbitrary
Depends on
Required by
Log
- 2026-04-28T21:56:20.174722167+00:00 Lightweight assignment: agent=Default Assigner (eea940a6), exec_mode=full, context_scope=full, reason=Default Assigner has 52 prior assignment tasks and domain expertise in agent assignment; this is fundamentally an assignment/evaluation task requiring CLI access and state modification.
- 2026-04-28T21:56:23.805485125+00:00 Spawned by coordinator --executor claude --model opus
- 2026-04-28T21:56:34.458208867+00:00 Starting quality pass for diagnose-tui-new and fix-tui-new
- 2026-04-28T21:57:33.874922740+00:00 diagnose-tui-new=research; agency stats: best_role=Programmer(52335de1) score=0.835 count=2, best_model=opus. Assigning Careful Programmer (f5143935), model=opus
- 2026-04-28T21:59:03.734799358+00:00 Quality pass complete. diagnose-tui-new + fix-tui-new both assigned to Careful Programmer (f5143935), model=claude:opus, status=open
- 2026-04-28T21:59:08.386259335+00:00 Task pending eval (agent reported done; awaiting `.evaluate-*` to score)
- 2026-04-28T22:02:10.583673884+00:00 PendingEval → Done (evaluator passed; downstream unblocks)