Wave Execute — Optimal-Model Bulk Hub Authoring
Orchestrates any wave of the OpenClaw vault hub system using the right model for each task.
Model selection matrix
| Task type | Model | Cost/M |
|---|---|---|
| Bulk templated hub generation | Gemma 4 26B | $0.06 |
| Science/reasoning hubs | Gemma 4 31B | $0.13 |
| Audit checks | Claude Haiku | $0.25 |
| Novel hub type / first instance | Claude Sonnet | $3.00 |
| Script / webhook / cron pages | Kimi | ~$0.15 |
| Orchestration / planning | Claude Opus | $15.00 |
Execution protocol
Step 0 — Model selection gate
Before doing any other work, invoke /model-selector to confirm model for each task category.
Step 1 — Load wave spec
Step 2 — Pre-flight (idempotency check)
Step 3 — Context compilation
Step 4 — Gemma generation
Step 5 — Write output to BOTH canonical + vault paths
Step 6 — Audit (AU-A, AU-B, AU-C)
Step 7 — Morning brief + close
Invokes / Invoked by
Invokes: SKILL, SKILL Invoked by: _summary, _summary
Recent activity
- 2026-05-04: skill created