Rediscover
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OMP_NUM_THREADS | No | Environment variable shown in the MCP server configuration to limit OpenMP threads, e.g. "2". | |
| OPENBLAS_NUM_THREADS | No | Environment variable shown in the MCP server configuration to limit OpenBLAS threads, e.g. "2". |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
Use tool outputs for every measurement. Never claim discovery significance or Bayesian mass uncertainties. Start with get_guide. Reports are generated from evidence, not model prose.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| get_guideB | Get the workflow, physics scope, privacy and interpretation limits. |
| list_eventsC | List one page of the official GWOSC catalog; follow has_more explicitly. |
| fetch_strainC | Download a bounded official H1/L1 strain pair and create an immutable run. |
| check_data_qualityB | Check finite samples, signal quality and GWOSC flags; returns evidence ID. |
| analyze_detectorC | PSD-weighted bounded template-bank search. Coarse detector-frame masses, no significance. |
| get_resultsB | Return computed detector measurements and cross-detector timing with evidence IDs. |
| get_catalogC | Fetch published inference AFTER analysis. Source-frame masses are not detector-frame masses. |
| make_reportB | Render raw/whitened strain and Q-transform plots plus an offline HTML evidence report. |
| verify_artifactsA | Verify saved artifact checksums; this checks integrity, not scientific correctness. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 9 tools
Each tool maps to a distinct pipeline stage (orient, list, fetch, quality-check, analyze, retrieve, report, verify), and descriptions explicitly warn about the main confusion points (detector-frame vs source-frame, get_results vs get_catalog). The only residual overlap is between list_events and get_catalog, both catalog-facing, plus three 'get_*' tools that an agent must read carefully to distinguish.
All nine names are snake_case verb_noun constructs (get_guide, fetch_strain, analyze_detector, make_report) with a predictable verb-first pattern. Verbs vary appropriately by action rather than chaotically, and resource nouns are clear.
Nine tools is well within the ideal 3-15 range and each corresponds to a real step in the gravitational-wave analysis workflow. No tool appears redundant or bolted on; even verify_artifacts and get_guide earn their place as integrity and orientation steps.
The surface covers the full lifecycle: orientation, catalog listing, strain acquisition, quality gating, analysis, results, published inference, reporting, and artifact verification. Minor gaps exist — no run cleanup/deletion, no cross-event comparison, and no direct export of results beyond the HTML report — but these are workaroundable.