Belt stalled: only 4 active learnings (10 required for yon-kyu)

[cmbays]

Observation

After Keiko 10, the belt calculator reports 4 active learnings against a yon-kyu threshold of 10.

Root cause (confirmed): 7 learning files exist on disk, 3 are correctly archived as Keiko 4 duplicates (95f2ade3, cb2c6914, efcde370). The archive filter is working correctly — this is not a bug.

Gap: Synthesis and extraction pipelines have not generated enough learnings yet. With 4 active learnings across ~10 completed cycles, the rate is ~0.4 learnings/cycle.

Impact

Belt stuck at go-kyu. Yon-kyu requires learningsTotal >= 10.

What needs to happen

1. Fix observations flowing into synthesis (closed in Keiko 9 bug: synthesis input always 0 observations — intelligence loop has no signal #335, verify it works now)
2. Ensure cooldown synthesis creates learnings per cycle — review LearningExtractor pipeline
3. Consider lowering yon-kyu threshold or documenting expected ramp-up time

[coderabbitai]

🔗 Related PRs

#339 - fix: keiko-9 batch — stale dist, ranWithYolo, bet.runId backfill, 0 observations [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!