INTERFACES¶
Commands, endpoints, and payloads for interacting with RUNE.
CLI Commands¶
python -m rune provides the following primary commands:
run-ollama-instance: Provision or select an Ollama server.run-agentic-agent: Run HolmesGPT analysis.run-benchmark: Multi-phase workflow for provisioning + analysis.vastai-list-models: List configured Vast.ai model catalog.ollama-list-models: List models exposed by an existing Ollama server.
CLI Options Summary¶
Shared Agent Options¶
--question,-q: The analysis question for HolmesGPT.--model,-m: The LLM model name to use.--kubeconfig: Path to the Kubernetes config file.--ollama-warmup,--no-ollama-warmup: Whether to pre-load the model before use.
Vast.ai Options (with --vastai)¶
--vastai-template: The Vast.ai template hash.--vastai-min-dph,--vastai-max-dph: Dollars-per-hour range.--vastai-reliability: Minimum machine reliability percentage.
API Server (HTTP Mode)¶
Run the server with python -m rune.api.
Endpoints¶
POST /v1/jobs/agentic-agent: Submit an agent analysis job.POST /v1/jobs/benchmark: Submit a full benchmark job.POST /v1/jobs/llm-instance: Submit an LLM backend provisioning job.POST /v1/jobs/ollama-instance: Deprecated alias forllm-instance.GET /v1/jobs/{job_id}: Poll for job status and results.GET /v1/catalog/vastai-models: List available Vast.ai models.GET /v1/llm/models?backend_url=<url>&backend_type=ollama: List models from a target backend.GET /v1/ollama/models?backend_url=<url>: Deprecated alias for the backend-generic models endpoint.
Auth Headers¶
Authorization: Bearer <token>X-API-Key: <key>X-Tenant-ID: <tenant>
REST API Reference¶
For a formal definition of all endpoints and request/reponse schemas, see the API Specification.
Advanced Inspection Views¶
These routes are primarily surfaced by rune-ui, but they are backed by
first-class rune API contracts and are useful integration targets in their
own right.
| Surface | UI route | API route | Notes |
|---|---|---|---|
| Run detail | /runs/{run_id} |
GET /v1/jobs/{job_id} |
Status, result metadata, telemetry |
| Live trace | /runs/{run_id} |
GET /api/v1/runs/{run_id}/trace |
UI proxy for the run event stream |
| Browser live view | /runs/{run_id} |
GET /api/v1/runs/{run_id}/browser-stream |
UI proxy for browser screenshots |
| Chain DAG | /chains/{run_id} |
GET /v1/chains/{run_id}/state |
Multi-agent graph state (nodes, edges, overall_status) |
| Audit artifacts | /audits/{run_id} |
GET /v1/audits/{run_id}/artifacts |
Compliance evidence list grouped by artifact kind |
| Audit download | /audits/{run_id} |
GET /v1/audits/{run_id}/artifacts/{artifact_id} |
Streams raw artifact bytes |
See Advanced Pipelines for behavior and validation details.
Wire Protocol (Driver Layer)¶
HolmesGPT is invoked via the DriverTransport layer.
Protocol JSON¶
Example ask action payload:
{
"action": "ask",
"params": {
"question": "What's wrong?",
"model": "llama3.1:8b",
"kubeconfig": "..."
}
}
The driver returns a structured JSON response with analysis results.