Research MCP servers

Pre-build reality check for AI coding agents. Scans GitHub, Hacker News, npm, PyPI, and Product Hunt to detect existing competition before building, returning a reality signal score (0-100), duplicate likelihood, similar projects, and pivot hints.

Evidence-backed open-source AI project intelligence for agents. Nine read-only tools search a live Radar, inspect and compare projects, shortlist candidates, and plan constraint-aware stacks while keeping verified facts, recommendations, unknowns, and risks separate. Hosted Streamable HTTP at `https://mcp.aiworkstation.cn/mcp` with no authentication, or install `aiworkstation-open-source-intelligence[mcp]==0.3.3` from PyPI. Public pre-release.

Query daily trending open-source projects from GitHub & Hacker News with AI-generated bilingual (EN/ZH) summaries, categories, use cases, and custom scoring. Stateless, no API key required.

Academic peer-review and research workflow tools from [Science AI Journal](https://scienceaijournal.com): Pre-Check (paper-tier prediction), AI multi-agent peer review, Research Gaps Finder, Journal Recommender (1,214 venues with predatory-journal flags), Duplicate Publication Checker (CrossRef/arXiv/medRxiv/bioRxiv/Unpaywall/900k-paper library), and the Article Writer pipeline. Free local tools (FTS5, zero LLM cost); LLM tools use caller account credits.

Full read/write access to a Zotero library — 39 tools covering items, collections, tags, attachments, notes, and bibliography/citation export. Self-service, multi-tenant: sign in with your own Zotero API key via a first-party OAuth 2.1 flow, no shared credentials. Version-gated writes, with an idempotency-key option to make retried deletes/moves safe.

Research workspace MCP integrating Zotero, Obsidian, and NotebookLM. Search papers (arXiv/Semantic Scholar/PubMed/CrossRef), ingest into Zotero, sync notes to Obsidian, verify NotebookLM briefs. All three external tools optional. `pip install research-hub-pipeline` then `research-hub serve`.

Deep research MCP server for OpenAI Responses API or Open Deep Research (smolagents), with web search and code interpreter support.

Search scientific papers with structured experimental data extracted from full-text studies. Returns 25+ fields per paper including methods, results, sample sizes, and quality scores. 50 free searches, then $0.01/result.

French election results from 1999 to 2026 at polling-station level (55 elections), cross-referenced with INSEE socio-demographic data, elected representatives and official administrative boundaries. Electoral maps, vote-reserve analysis, cross-election panels and read-only SQL over DuckDB. Hosted, no auth, no install: `https://mcp.andre.vote/mcp`, also listed as a [hosted connector on Glama](https://glama.ai/mcp/connectors/vote.andre/elections).

Self-hosted perception server for agents: local bilingual ASR, in-band vision, cross-lingual search, login-walled sources you hold credentials for, and a persistent relation graph.

Cinema-quality science visualization MCP server for CFD/FEA/SPH. 22 tools for rendering, slicing, contouring, volume rendering, and animating OpenFOAM/VTK/CGNS data. Headless EGL/OSMesa.

Search and fetch research datasets across Zenodo, DataCite (Dryad/Figshare/Dataverse/OSF), NCBI omics (GEO/SRA/BioProject), and literature (PubMed/OpenAIRE) behind one normalized model — DOI deduplication, NCBI-Taxonomy synonym expansion, paper→data linking, and checksum-verified download. `uvx data-aggregator-mcp`.

Zotero MCP server (`zotio-mcp`) with a capability-tagged tool registry and preview-first, journaled writes (an agent can't silently mutate or delete your library). Reads work keyless against the local Zotero desktop API; writes route to the Web API. Ships an MCPB bundle for one-click Claude Desktop install.

Query real, quality-scored humans as an agent tool. [GetABrain](https://getabrain.ai) routes a question to paid human workers and returns schema-validated JSON — 16 response types (yes/no, ratings, rankings, A/B tests, sentiment, image/video/audio review, voice/video/photo capture). Fully programmatic signup with a $5 free trial credit, no card required. `npx -y @getabrain/mcp-server`.

Natural-language patent and scientific literature search across 200M+ patents and 216M+ papers from 170+ jurisdictions, with semantic/BM25 search, structured filters, and record retrieval. Hosted Streamable HTTP endpoint and Dockerized stdio bridge; requires a free Patsnap API key.

MCP server for [ITASCA PFC](https://www.itascacg.com/software/pfc) discrete element simulation — browse documentation, execute scripts, capture plots, and manage long-running tasks via a WebSocket bridge to the PFC GUI.

Patent prior art search and startup IP intelligence for AI founders. Clone risk score, defensibility map, patentable features, competitor patent landscape — free analysis in ~2 minutes. Full IP Portfolio Package (6 documents) available for $99. `pip install fittin-mcp`

Philippine government data as agent-callable tools: PHIVOLCS earthquakes + volcano alerts, PAGASA weather + typhoons, PhilGEPS procurement, PSA 2020 Census population + poverty, AQICN air quality. Install: `uvx ph-civic-data-mcp`.

Provenance-first MCP server for the UNESCO Institute for Statistics (UIS): ~5,000 indicators on education (enrolment, completion, literacy, expenditure), science/R&D, culture and communication by country, region and year, with source URL, data release, retrieval timestamp and licence on every response. Hosted or local via `npx uis-mcp-server`.

Full-text search of the Library of Congress's digitised American newspapers, plus catalogue search per media type and the curated collections. No API key. `npx -y mcp-libraryofcongress`

AI-analyzed startup & product opportunities mined from Reddit, Hacker News, Product Hunt & Stack Exchange. Query pain points, trending themes, and commercial scores (pain intensity, willingness-to-pay, build difficulty). Remote hosted MCP at painspotter.ai — free tier covers overview + opportunity detail; Pro unlocks search, trending themes & theme-level market signals.

Send research papers from Claude to a Kindle or Kobo, with Zotero as the source of truth. Searches arXiv and OpenAlex, recommends related work via the Semantic Scholar citation graph, queues papers in a Zotero collection, and delivers the PDFs by email or Dropbox. Receipts report file sizes and name anything that could not be sent.

MCP server for patent search and prior art discovery powered by Google Patents public dataset on BigQuery. Tools: search patents, get full patent details with CPC codes and citations, fetch legal claims text. Free to use.

Semantic Scholar MCP server: 14 typed tools over 200M+ academic papers — search, paper/author retrieval, citation-graph traversal, and recommendations. `uvx s2-mcp-server`

Remote MCP server for Google Scholar: search with author and year filters, cited-by and all-versions lookups, and citations in five styles with BibTeX export, as JSON.

Search, cite, and compare the Pāli Canon (Tipiṭaka) — full coverage at parity with SuttaCentral (~444K segments across Sutta, Vinaya, Abhidhamma). Hybrid trigram+vector search, full-sutta fetch with cross-reference URLs (SuttaCentral + 84000.org), segment-aligned translation comparison, and Pāli word lookup against the P. A. Payutto dictionary. Free, non-commercial — offered as Dhamma Dāna. Hosted at `https://tripitaka-mcp.com` or self-host via Docker.

Read-only MCP server and live-map dataset for declassified U.S. federal UAP/UFO records, with searchable records, hotspots, timelines and direct source links.

Full-text search inside the Internet Archive's scanned books, periodicals and documents, catalogue and book search, and Wayback Machine captures of a URL at a given date. No API key. `npx -y mcp-archiveorg`

MCP server for [YADE](https://yade-dem.org/) — open-source discrete element method (DEM) engine for granular and particle simulation. Browse API docs with BM25 search, execute code via synchronous REPL or async background tasks, monitor progress, and review task history.

Instantly understand what real users think and why by querying our network of 1:1 Digital Twins - each representing a real person. Give your AI agents authentic human context to ground outputs, improve creative, and make smarter decisions.

Resolve any scholarly identifier (DOI, PMID, PMCID, ISBN, arXiv, ISSN, ADS bibcode, WHO IRIS URL) into structured CSL JSON, format in 10,000+ citation styles, or export to BibTeX/RIS/EndNote/Zotero RDF/CSV. Single or batch. Install: `npx -y scholar-sidekick-mcp`.

Remote MCP server for Google Trends: interest over time, interest by region, and rising and top related queries and topics for any keyword, as JSON.

Hourly LLM-ranked news briefings (AI, frontier labs, markets, sports, world) as structured JSON, rendered 480x800 e-ink pages and a free provenance dump of every candidate with dedupe metadata. 25 free calls/day, then a $9/month key or x402. Install via `npx -y github:jshelley/briefing-mcp`.

Search and cite rare historical texts (alchemy, Hermeticism, Renaissance philosophy) with DOI-backed academic citations from [Source Library](https://sourcelibrary.org)

Auditable strategic-risk memos for AI agents — sanctions, regulatory, geopolitics, trade. Geography-routed regional specialists (Central Asia/Caspian, Gulf/Middle East), per-claim evidence discipline, schema-validated output, machine-verified audit. No live data retrieval. PyPI: `agenda-intelligence-md`.

Search 474 validated startup gaps from your AI editor. Find unsolved business problems by industry, role, or keyword — each with pain point, solution direction, and link to full blueprint. Zero API keys, works offline.

MCP server for [SPARKIT](https://sparkit.science), a scientific research agent that searches the literature, reads papers, and returns cited Markdown reports. Submit a question via the `research` tool from Claude Desktop / Cursor / Claude Code; get back a ~90-second cited report without leaving the chat. `uv tool install sparkit-mcp` or `pip install sparkit-mcp`.

Freedom-of-information requests for Germany via [FragDenStaat.de](https://fragdenstaat.de), the FOI portal built on [froide](https://github.com/okfde/froide). Finds the responsible authority from a place name and returns the evidence trail, reads the applicable act and its statutory deadline, tracks your own requests and their replies, and drafts new requests as local YAML checked against 25 rules before anything is sent. Four tools need no account. Submitting is guarded by five independent gates and defaults to a dry run, because the API sends immediately and cannot be undone.

Conduct AI-powered qualitative research interviews and surveys at scale with [Agent Interviews](https://agentinterviews.com). Create interviewers, manage research projects, recruit participants, and analyze interview data through MCP.

Zotero MCP server in TypeScript: hybrid keyword and semantic search, PDF passages with page locators, bibliographies in any CSL style (citeproc-js), add by DOI, group libraries, and writes (items, tags, collections, attachments, PDF annotations anchored to the quoted passage). Reads and personal-library writes go to the running Zotero app without a cloud key. `npx -y @oscardvs/zoteus`, or a one-click `.mcpb` for Claude Desktop.

Check a citation against Crossref for retraction, correction, or expression-of-concern status before it goes into a paper, summary, or review — catches papers like Wakefield's retracted MMR-autism study, still cited 1,000+ times since its 2010 retraction. No API key required. Detection logic (checked against real Crossref data, not just documentation) is shared identically between a Python CLI/library, this MCP server, and a GitHub Action for CI. `npx -y citeguard-mcp-server`

Query 9 classical Chinese texts (pre-Qin to Wei-Jin: 史记/汉书/后汉书/三国志/左传/论语/孟子/吕氏春秋/资治通鉴) by event, person, modern place name, or virtue — every result cites 【book · chapter · paragraph】 with original-text evidence and an honest `review_status`. Zero-dependency stdlib stdio JSON-RPC.

MCP server for US immigration guidance — live Visa Bulletin, priority date checker and immigration term explanations. Free to use.

Evidence-grounded web research for AI agents. Verifies page-local claims with citations, confidence bands, policy reports, and multi-page session synthesis via MCP.

Access legislative data from all 50 US states and Congress — search bills, get full text, track votes, and look up legislators via the LegiScan API.

Read and write a local Zotero 10+ library through Zotero's own local API: collections, items, attachments from disk (linked or imported), indexed full text, BibTeX and CSL export, tags and saved searches. Deleting goes to Zotero's trash and is restorable. No Zotero plugin, no zotero.org API key and no cloud round trip, so it works offline. Install: `npx -y zotero-native-mcp`

Register LLMs/agents and submit research papers (Markdown) to the [BenchClaw](https://www.p2pclaw.com/app/benchmark) leaderboard. Papers are scored by a 17-judge Tribunal with 8 deception detectors across 10 dimensions. No API key required. Works with Claude Desktop, Cursor, Cline, Zed, Continue.dev.

Chemistry toolbox computed from authoritative data: molecular weight / molar mass of any formula (nested groups, isotope labels, propagated NIST uncertainties, percent composition), isotope-distribution mass spectra (m/z peaks for a given charge), and name⇆formula lookup with isomer resolution. Element masses from the NIST Atomic Weights and Isotopic Compositions database; offline and deterministic, with an optional on-by-default online fallback (PubChem/Wikidata/EPA CompTox) that can be disabled. `uvx mcp-molecules`.

Independent trust scores, claim audits, and comparisons for AI agents — queryable by your agent over MCP. Hosted Cloudflare Worker at hlido.eu/mcp (no install). Returns a 0–100 score, tier verdict, per-claim PASS/FAIL audit, and signed evidence for a reviewed agent. From [Hlido](https://hlido.eu).

Official AgentFEM bridge for AI agents to create, validate, run, inspect, and verify finite-element simulations through seven typed tools. Install: `uvx --from agentfem-mcp agentfem-mcp`.

Semantic search over 600k+ CS/AI papers with citation-graph traversal, full-text extraction, embeddings, and BibTeX export. Works anonymously or with a free key. Install: `npx scholar-feed-mcp init`.

Evidence-backed web research for AI agents. Real-time search with cited claims, confidence scores, and compare mode (raw LLM vs evidence-backed). MCP server, REST API, and Python SDK.