Quantum Research Hub MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| QRH_DB_PATH | No | Path to the SQLite database file. | |
| QRH_DATA_DIR | No | Directory for data artifacts. | |
| QRH_CLAUDE_MODEL | No | Claude model name for model pass (e.g., claude-sonnet-4-5). | |
| ANTHROPIC_API_KEY | No | Anthropic API key for Claude model pass. | |
| QRH_LOOKBACK_DAYS | No | Number of days to look back for arXiv papers. | |
| QRH_BUDGET_PROFILE | No | Budget profile: low, medium, or high. | |
| QRH_MEMORY_BACKEND | No | Memory backend: 'bm25' or 'hybrid'. | |
| QRH_APPROVAL_GRANTED | No | Set to '1' to skip approval gating for experiments. | |
| QRH_ENABLE_MODEL_PASS | No | Set to '1' to enable optional Claude model pass. | |
| QRH_ARXIV_MIN_INTERVAL | No | Minimum interval in hours between arXiv fetches. | |
| QRH_EXPERIMENT_TIMEOUT_SECONDS | No | Timeout for experiment runs in seconds. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
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
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| search_arxivC | Search arXiv for quantum-computing papers by query, categories, keywords, and date range. |
| ingest_paperC | Ingest one arXiv paper: metadata -> chunks -> deterministic paper card -> DB. |
| get_paper_cardC | Return the compact paper card (scores, methods, claims, possible experiments). |
| search_paper_memoryC | Search stored paper memory with the configured local retrieval backend. |
| list_recent_papersB | List recently-seen papers, optionally filtered by minimum relevance score. |
| create_ideaC | Create a research idea. MUST cite >=1 source arXiv paper (rejected otherwise). |
| list_ideasC | List research ideas, optionally filtered by status. |
| create_experiment_from_ideaC | Create a reproducible experiment folder from an idea (baseline + tests + config + metrics). |
| get_experimentA | Return full experiment detail: metadata, config, latest metrics, validator notes. |
| list_experimentsC | List experiments with latest metric + validator verdict. |
| run_experimentA | Run an experiment safely (small=autonomous; gpu/medium/long need approval). Timeout-bounded. |
| get_experiment_resultsB | Return the latest run's status, metrics, and log path for an experiment. |
| validate_experimentC | Skeptically validate an experiment; verdict accepted|rejected|inconclusive. |
| create_daily_reportA | Generate the daily research report markdown for a date (default: today). |
| create_weekly_reportA | Generate the weekly research report markdown (week_start normalized to Monday). |
| get_budget_statusB | Return the active budget profile, daily caps, and usage so far today. |
| get_overviewB | Return dashboard headline counts, recent activity, and budget status. |
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 17 tools
Most tools have distinct purposes, but get_experiment and get_experiment_results overlap in returning experiment data; an agent might confuse which one to use. Other tools are clearly separated.
All tools follow a consistent verb_noun pattern (create_, get_, list_, etc.), making it predictable and easy to understand the action and target.
With 17 tools, the set covers the primary research workflow (paper ingestion, idea creation, experiment lifecycle, reporting, budget) without being overwhelming or sparse.
The core workflow is well-covered, but missing update/delete operations for experiments, ideas, and papers could cause dead ends if agents need to modify or remove resources.