npx claudepluginhub supportersimulator/3-surgeons --plugin 3-surgeonsThis skill uses the workspace's default tool permissions.
Probe is a health check -- it pings both the Cardiologist and Neurologist endpoints to verify they are reachable and responding. Run it:
Generates design tokens/docs from CSS/Tailwind/styled-components codebases, audits visual consistency across 10 dimensions, detects AI slop in UI.
Records polished WebM UI demo videos of web apps using Playwright with cursor overlay, natural pacing, and three-phase scripting. Activates for demo, walkthrough, screen recording, or tutorial requests.
Delivers idiomatic Kotlin patterns for null safety, immutability, sealed classes, coroutines, Flows, extensions, DSL builders, and Gradle DSL. Use when writing, reviewing, refactoring, or designing Kotlin code.
Probe is a health check -- it pings both the Cardiologist and Neurologist endpoints to verify they are reachable and responding. Run it:
Do NOT use for: routine operations where partial degradation is acceptable, or as a substitute for gains-gate (probe checks connectivity only, not full system health).
3s probe
probe()
No arguments needed. Probe checks all configured surgeons.
Probing surgeons...
Cardiologist: OK (245ms)
Neurologist: OK (89ms)
Atlas (Claude): always available (this session)
All surgeons operational.
{
"cardiologist": {"status": "ok", "latency_ms": 245},
"neurologist": {"status": "ok", "latency_ms": 89},
"atlas": {"status": "ok", "note": "always available (this session)"}
}
| Status | Meaning |
|---|---|
ok | Surgeon responded successfully. latency_ms shows response time. |
fail | Surgeon responded but with an error (e.g., invalid API key, model not found). Check the error field. |
unreachable | Could not connect to the endpoint at all (network issue, service not running). |
Atlas (Claude) is always reported as ok because it IS the current session. There is no endpoint to ping -- if you are reading this output, Atlas is operational.
| Symptom | Likely Cause | Fix |
|---|---|---|
unreachable | Network issue, OpenAI outage | Check internet, check status.openai.com |
fail with 401 | Invalid API key | Verify OPENAI_API_KEY env var is set and valid |
fail with 429 | Rate limited | Wait, reduce request frequency |
fail with 500+ | OpenAI server error | Retry in a few minutes |
Impact of Cardiologist being down: Cross-examination runs with Neurologist only. Consensus uses only Neurologist's vote (weighted score less reliable). A/B test proposals still work (local). Cost tracking pauses for external calls.
| Symptom | Likely Cause | Fix |
|---|---|---|
unreachable | Ollama/MLX not running | Start Ollama: ollama serve |
unreachable | Wrong endpoint in config | Check ~/.3surgeons/config.yaml neurologist endpoint |
fail | Model not pulled | Pull the model: ollama pull qwen3:4b |
Impact of Neurologist being down: Cross-examination runs with Cardiologist only. Consensus uses only Cardiologist's vote. Local pattern recognition unavailable. All operations still function but with reduced diversity of perspective.
If both external surgeons are unreachable, Atlas (Claude) is still operational but without the multi-model consensus benefit. In this state:
3s init to reconfigure if endpoints changed| Probe | Gains-Gate | |
|---|---|---|
| Checks | Surgeon connectivity only | Connectivity + evidence store + state backend + more |
| Speed | Fast (~500ms) | Medium (~2-5s) |
| Use case | Quick health check | Phase transition verification |
| Blocks on failure | No (informational) | Yes (critical checks block progress) |
Use probe for quick checks. Use gains-gate between major phases when you need to verify the full system is healthy.
Healthy response times to expect:
| Surgeon | Typical Latency | Concerning |
|---|---|---|
| Cardiologist (OpenAI) | 200-500ms | >2000ms |
| Neurologist (local Ollama) | 50-200ms | >1000ms |
High latency on the Neurologist usually indicates the local model is loading into memory (first request after start). Subsequent requests should be faster.