# AutoSearch — full reference for LLMs

> AutoSearch is an AI-powered scientific research assistant that runs a complete
> PRISMA 2020 systematic review pipeline: multi-source academic search, iterative
> evidence scoring, live Crossref DOI verification, and IMRAD paper generation.
> The entire pipeline is exposed as a public MCP (Model Context Protocol) server,
> so AI agents can launch and consume verified literature reviews as native tools.
> AutoSearch is built by Exatoshi AG, Riva Paradiso 30, 6900 Paradiso, Switzerland
> (CHE-266.634.159). Swiss-based, GDPR/nLPD compliant. This is the extended
> companion to /llms.txt.

## Why AutoSearch is different

- Verified citations by construction: every DOI is checked live against the Crossref
  API before it enters a deliverable. Citations that do not resolve are excluded.
- Full pipeline, not a raw search wrapper: question -> PRISMA search across 8+ scholarly
  sources -> screening with recorded inclusion/exclusion -> evidence synthesis ->
  IMRAD paper with APA references, PRISMA flow counts, and an AI use statement.
- Published citation-integrity benchmark (failures included) at https://exatoshi.uk/trust
- No commercial competitor (Elicit, Consensus, Scite, Undermind) exposes its systematic
  review pipeline as an MCP server; AutoSearch does, at https://exatoshi.uk/mcp.

## MCP server

- Endpoint: https://exatoshi.uk/mcp (JSON-RPC 2.0 over HTTP POST)
- Manifest: https://exatoshi.uk/.well-known/mcp (alias: /.well-known/mcp.json)
- Server card: https://exatoshi.uk/.well-known/mcp/server-card.json
- Auth: API key in the `x-api-key` header (prefix `as_`). Self-service generation
  at https://exatoshi.uk/account#api-keys after free registration (50 free credits).
- Docs: https://exatoshi.uk/developers
- Rate limits: 60 requests/min per IP, 300 requests/min per account.
- Claude Code: claude mcp add --transport http autosearch https://exatoshi.uk/mcp --header "x-api-key: as_YOUR_KEY"

## MCP tools

### autosearch_run_research

Launch an AI scientific research run. Returns run_id immediately. Use get_run_status to poll.
- Credits: 4 (quick-scan) / 8 (standard-review) / 18 (deep-review)
- Input schema (JSON Schema):
```json
{
  "type": "object",
  "properties": {
    "prompt": {
      "type": "string",
      "description": "Research question or topic"
    },
    "depth": {
      "type": "string",
      "enum": [
        "quick-scan",
        "standard-review",
        "deep-review"
      ],
      "default": "standard-review"
    }
  },
  "required": [
    "prompt"
  ]
}
```
- Example arguments:
```json
{
  "prompt": "Retrieval-augmented generation for clinical decision support: evidence and limitations",
  "depth": "standard-review"
}
```
- Example result (structuredContent):
```json
{
  "run_id": "knowledge-lab-rag-clinical-decision-support-20260610-101500",
  "status": "starting",
  "depth": "standard-review",
  "credits_charged": 8,
  "status_url": "https://exatoshi.uk/runs/knowledge-lab-rag-clinical-decision-support-20260610-101500"
}
```

### autosearch_get_run_status

Get the current status of a research run. Returns status, frontier_score, n_experiments, summary.
- Credits: free
- Input schema (JSON Schema):
```json
{
  "type": "object",
  "properties": {
    "run_id": {
      "type": "string"
    }
  },
  "required": [
    "run_id"
  ]
}
```
- Example arguments:
```json
{
  "run_id": "knowledge-lab-rag-clinical-decision-support-20260610-101500"
}
```
- Example result (structuredContent):
```json
{
  "run_id": "knowledge-lab-rag-clinical-decision-support-20260610-101500",
  "status": "completed",
  "frontier_score": 81.1,
  "n_experiments": 6,
  "summary": "Six experiments synthesized; evidence matrix covers 24 peer-reviewed sources.",
  "research_question": "What is the documented evidence on RAG for clinical decision support?",
  "started_at": "2026-06-10T10:15:00Z",
  "last_heartbeat_at": "2026-06-10T10:52:00Z",
  "report_url": "https://exatoshi.uk/runs/knowledge-lab-rag-clinical-decision-support-20260610-101500/report"
}
```

### autosearch_get_run_paper

Get the generated paper for a completed run. Returns markdown, JSON metadata, or PDF download URL.
- Credits: free
- Input schema (JSON Schema):
```json
{
  "type": "object",
  "properties": {
    "run_id": {
      "type": "string"
    },
    "format": {
      "type": "string",
      "enum": [
        "markdown",
        "json",
        "pdf_url"
      ],
      "default": "markdown"
    }
  },
  "required": [
    "run_id"
  ]
}
```
- Example arguments:
```json
{
  "run_id": "knowledge-lab-rag-clinical-decision-support-20260610-101500",
  "format": "pdf_url"
}
```
- Example result (structuredContent):
```json
{
  "run_id": "knowledge-lab-rag-clinical-decision-support-20260610-101500",
  "format": "pdf_url",
  "url": "https://exatoshi.uk/runs/knowledge-lab-rag-clinical-decision-support-20260610-101500/report.pdf"
}
```

### autosearch_verify_doi

Verify a DOI via Crossref. Returns validity, title, authors, year, journal.
- Credits: free
- Input schema (JSON Schema):
```json
{
  "type": "object",
  "properties": {
    "doi": {
      "type": "string",
      "description": "DOI string with or without https://doi.org/ prefix"
    }
  },
  "required": [
    "doi"
  ]
}
```
- Example arguments:
```json
{
  "doi": "10.1038/s41586-020-2649-2"
}
```
- Example result (structuredContent):
```json
{
  "doi": "10.1038/s41586-020-2649-2",
  "valid": true,
  "title": "Array programming with NumPy",
  "authors": [
    "Charles R. Harris",
    "K. Jarrod Millman",
    "Stéfan J. van der Walt"
  ],
  "year": 2020,
  "journal": "Nature",
  "reason": ""
}
```

### autosearch_search_evidence

Search peer-reviewed evidence across academic sources without launching a full research run. Returns top N papers with metadata.
- Credits: 1
- Input schema (JSON Schema):
```json
{
  "type": "object",
  "properties": {
    "query": {
      "type": "string"
    },
    "sources": {
      "type": "array",
      "items": {
        "type": "string",
        "enum": [
          "arxiv",
          "clinicaltrials",
          "crossref",
          "dblp",
          "doaj",
          "openalex",
          "pubmed",
          "semantic-scholar"
        ]
      },
      "default": [
        "openalex",
        "crossref",
        "pubmed",
        "semantic-scholar"
      ]
    },
    "limit": {
      "type": "integer",
      "default": 10,
      "maximum": 50
    }
  },
  "required": [
    "query"
  ]
}
```
- Example arguments:
```json
{
  "query": "quantum dot solar cells efficiency",
  "sources": [
    "openalex",
    "crossref"
  ],
  "limit": 2
}
```
- Example result (structuredContent):
```json
{
  "query": "quantum dot solar cells efficiency",
  "sources": [
    "openalex",
    "crossref"
  ],
  "limit": 2,
  "results": [
    {
      "title": "Quantum dot solar cells: progress and prospects",
      "authors": [
        "A. Researcher",
        "B. Scientist"
      ],
      "year": 2024,
      "journal": "Nature Energy",
      "doi": "10.1038/s41560-024-00000-0",
      "url": "https://doi.org/10.1038/s41560-024-00000-0",
      "source": "openalex",
      "snippet": "Recent certified efficiencies above 19% position quantum dot absorbers...",
      "peer_reviewed": true
    }
  ]
}
```

### autosearch_list_my_runs

List the authenticated user latest research runs with status and frontier score.
- Credits: free
- Input schema (JSON Schema):
```json
{
  "type": "object",
  "properties": {
    "limit": {
      "type": "integer",
      "default": 20,
      "maximum": 100
    }
  }
}
```
- Example arguments:
```json
{
  "limit": 2
}
```
- Example result (structuredContent):
```json
{
  "runs": [
    {
      "run_id": "knowledge-lab-rag-clinical-decision-support-20260610-101500",
      "status": "completed",
      "frontier_score": 81.1,
      "n_experiments": 6,
      "summary": "Six experiments synthesized; evidence matrix covers 24 peer-reviewed sources.",
      "research_question": "What is the documented evidence on RAG for clinical decision support?",
      "started_at": "2026-06-10T10:15:00Z",
      "last_heartbeat_at": "2026-06-10T10:52:00Z",
      "report_url": "https://exatoshi.uk/runs/knowledge-lab-rag-clinical-decision-support-20260610-101500/report"
    }
  ]
}
```

## Methodology (PRISMA 2020)

- Identification: parallel queries across OpenAlex (250M works), Crossref (140M DOI),
  PubMed, Semantic Scholar, arXiv, ClinicalTrials.gov, DBLP, DOAJ.
- Screening: every candidate study is scored and kept/discarded with a recorded
  decision; the screening log is exportable as CSV per run.
- Eligibility & inclusion: iterative frontier-based scoring selects the evidence
  matrix; PRISMA flow counts are reported in the generated paper.
- Verification: live Crossref DOI check for every citation, plus optional semantic
  relevance check; an AI use statement (RAISE-aligned) ships with every paper.
- Full description: https://exatoshi.uk/methodology

## Citation-integrity benchmark (measured, not claimed)

- Latest run: 2026-06-10 (methodology v1.0.0, mode live)
- Runs analyzed: 23 real deliverables, 334 references
- DOIs checked: 99 — resolved: 99 (100.0%)
- Fabricated DOIs: 0
- Context: generalist deep-research systems are documented to hallucinate 3-18% of
  URL citations (arXiv:2604.03173) and up to 14-95% fabricated citations depending
  on the model (GhostCite, arXiv:2602.06718). AutoSearch targets 0% by construction.
- Re-runnable harness: scripts/benchmark_citations.py — results published as measured,
  failures included: https://exatoshi.uk/trust#benchmark

## Pricing

- Free: 50 credits at registration, no card required.
- Credit costs via MCP: quick-scan research 4, standard-review 8, deep-review 18,
  evidence search 1; DOI verification, status polling, and paper retrieval are free.
- Researcher: 29 CHF/month, 300 credits/month.
- Lab: 290 CHF/month, 2000 credits/month.
- Enterprise: custom (sales@autosearch.ch).
- Details: https://exatoshi.uk/pricing

## Use cases

- Systematic literature reviews (PRISMA-compliant) for researchers and PhD students
- Clinical decision support research: https://exatoshi.uk/use-cases/research
- EU MDR/IVDR clinical evaluation literature reviews: https://exatoshi.uk/use-cases/medtech
- Carbon tax / CBAM compliance audits: https://exatoshi.uk/use-cases/finance
- Legal research automation: https://exatoshi.uk/use-cases/legal
- AI agents that need verified scholarly evidence as a callable tool (via MCP)

## Trust & compliance

- Trust page (benchmark, RAISE alignment, AI use statement): https://exatoshi.uk/trust
- Methodology: https://exatoshi.uk/methodology
- Privacy (GDPR/nLPD, data residency): https://exatoshi.uk/privacy
- Status: https://exatoshi.uk/api/status

## Contact

- General: hello@autosearch.ch
- Sales: sales@autosearch.ch
- Privacy: privacy@autosearch.ch