| name | kalibr-router | |||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| description | Route OpenClaw agents to cheaper models automatically. Reduces token burn, handles model failures, avoids rate limits, and auto-selects the best model based on real production outcomes. Uses Thompson Sampling. Zero config. 3-line install. | |||||||||||||||||||||||||
| version | 1.0.0 | |||||||||||||||||||||||||
| metadata |
|
Route OpenClaw agents to cheaper, faster, or more reliable models automatically — based on real production outcomes.
use cheaper models · reduce token burn · route to best model automatically · handle model failures automatically · auto select model based on performance · avoid rate limits by switching models · model failover · self-healing agents · token optimization · llm routing · openclaw plugin · agent reliability · cost reduction · model selection
Installs and configures the @kalibr/openclaw plugin. After setup, every agent run reports its outcome (model used, tokens, success/failure) to Kalibr. Kalibr uses Thompson Sampling to route subsequent runs to the best-performing execution path.
Use cases:
- Heartbeat token reduction: lighter tasks route to cheaper models automatically
- Autonomous model selection: no hardcoded models, Kalibr learns your workload
- Self-healing agents: detects rate limits and provider outages, reroutes before users notice
- Agent run observability: zero-config telemetry on every LLM call
openclaw plugins install @kalibr/openclaw
openclaw config set plugins.entries.kalibr.config.apiKey "YOUR_KEY"
openclaw gateway restartGet your API key: https://dashboard.kalibr.systems
openclaw plugins list
/kalibr1,000 traces/month. No credit card. https://dashboard.kalibr.systems/settings
- Kalibr: https://kalibr.systems
- Docs: https://kalibr.systems/docs
- GitHub: https://github.com/kalibr-ai/openclaw-kalibr
- Dashboard: https://dashboard.kalibr.systems