Log inSubmit server
ResearchZosho logo

ResearchZosho

Research questions across web, scholarly, archival, and personal sources with claim-level citations and reviewable evidence.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install ResearchZosho

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "researchzosho": {
      "command": "npx",
      "args": [
        "-y",
        "@wyrdsekai/researchzosho-mcp"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Automated check passed— started and listed 49 tools correctly (7d ago).

Overview

ResearchZosho MCP server connects AI hosts to a research library that searches the web, scholarly literature, archives, videos, and user-provided documents. It records claims with quotations and sources, checks cited sentences, and marks unresolved or inaccessible evidence. Researchers can query existing findings, start new investigations, review claims, and continue work from a local model or hosted API. Reach for it when traceable research matters more than a one-off generated answer.

Use cases

  • Investigate a question across web and scholarly sources
  • Analyze a folder of PDFs without starting with web search
  • Audit claims against quotations and source records
  • Track ongoing research and newer paper versions
  • Explore a personal book collection for related reading

Key features

  • Claim-level citations with source quotations
  • Web, scholarly, archival, video, and document research
  • Persistent library with accepted and disputed claims
  • Local or hosted language-model support
  • Browser pages, CLI commands, and MCP access
  • Scheduled searches and source-change monitoring

Tools49

What ResearchZosho lets your agent do. Expand a tool for its parameters.

Verified live~19.7k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 1d ago

Recent health check succeeded.

9/9 checks healthy over the last 37d

What it can do

From the 49 tools this server returned when we ran it.

Runs commands · 2Can delete or modify data · 6Handles credentials · 4Makes network requests · 13Read-only tools · 41
Network at startup5 connections
Files written at startup5
Install scriptsNone
npm provenanceSigned build
Package age53 days
Maintainers1

Observed 2026-11-04 in an isolated sandbox. A description of reach, not a security audit.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 14d ago via OSV.dev · @wyrdsekai/researchzosho-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars1Stargazers on the source repository.
npm downloads2kPackage downloads in the last 30 days.
Last commit2d agoMost recent push to the default branch.
Availability100%Our rolling endpoint + install checks that succeeded.
Install checkPassedOur sandbox started it and listed its tools.
Tools exposed49Callable tools this server registers over MCP.

How ResearchZosho works

What ResearchZosho does

ResearchZosho is an open-source research assistant built around a persistent library of sources, claims, and evidence. It accepts a question, searches relevant material, and produces a structured report in which claims point back to sources. The system can work across general web pages, scholarly papers identified through Crossref and OpenAlex, archived pages, historical newspapers, Japanese digitised books, source-code archives, video sites, and documents supplied by the user.

It also supports research that begins with a private collection. PDFs, web pages, and folders can be added as collections, after which questions can search only those materials or use them as the starting point for broader web research. Research can include sources in multiple languages, with the resulting report identifying the languages used.

How it works

A research run reads sources in parallel and continues until the task is complete or its allotted time ends. Findings are stored as individual claims rather than only as prose. Each claim records its source and a quotation, while the report includes a source list and an evidence table. Sentences that cite sources are checked against those sources before the write-up is shelved.

Claims begin as drafts. A user can accept, dispute, or retire them, and accepted findings can inform later questions. Single-source claims remain drafts until independently supported. Retracted papers can dispute claims that rely on them, banned sites can be excluded, and trusted sites can be treated as primary sources. Sources blocked by a paywall, login, or other access problem are identified so their text can be supplied separately.

The library can run scheduled searches, revisit accepted claims, detect newer paper versions, retain dated backups, and report possible duplicate claims or documents for cleanup. It does not merge or delete material automatically.

Setup and configuration

ResearchZosho runs on the user's computer and can use a local model through llama.cpp, Ollama, or LM Studio, or a hosted model API. An AWS Bedrock option supports models in the user's own AWS account with that account's sign-in. The project includes model-selection guidance based on local graphics memory, measured quality, runtime, and cost.

The service exposes browser pages for searching, asking questions, reading claims with their sources, browsing subjects, checking activity, and downloading reports as Markdown or PDF. The vault command creates a folder that Obsidian, SoloMD, or SilverBullet can import. The share command makes the library reachable outside the home behind encryption.

Tools and capabilities

  • Use researchzosho chat to converse with the Librarian, start research runs, ask follow-up questions, inspect the shelves, or process the inbox.
  • Use researchzosho ask to retrieve existing claims and their evidence for a question.
  • Use researchzosho myshelf like to find similar books in an imported Calibre library.
  • Use researchzosho map to explore connections among people, places, authors, and works.
  • Ask for book recommendations backed by library catalogues such as Open Library and Japan's National Diet Library.
  • Research purchases using current listings, prices, datasheets, recalls, repairability information, owner reports, and used-market links.
  • Investigate podcasts and videos with citations to episodes or timestamps, and follow shows for new-episode notices.
  • Let Claude Code, Codex, Gemini CLI, or another MCP host query the library directly.

Limitations and notes

ResearchZosho is designed for human review: nothing becomes valid until a person reviews it. Source access can fail because of paywalls, logins, unavailable pages, or other retrieval problems. Claims that lack independent support remain drafts, and unsupported figures are marked as guesses in conversation. Automated maintenance reports possible cleanup but leaves merging, deletion, and final decisions to the user.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about ResearchZosho

ResearchZosho is an MCP server that connects AI hosts to a persistent research library and its web, scholarly, archival, video, and personal-document sources. Its main tools support asking questions, starting research runs, reviewing claims, checking evidence, and exploring stored findings with citations and quotations.

Install and run ResearchZosho according to its project setup, then connect an MCP host to the library. It runs on your computer and can use llama.cpp, Ollama, LM Studio, a hosted model API, or AWS Bedrock.

Yes, Claude Code can query the ResearchZosho library directly through MCP. Codex, Gemini CLI, and other MCP hosts can also use it.

Yes, ResearchZosho can shelve PDFs, web pages, or an entire folder as a collection. Questions can search only those documents or begin with them and continue to the web.

Each finding records its source and a quotation, and reports include an evidence table and numbered source list. Source-citing sentences are checked against their sources before the report is shelved.

Related MCP servers

All in Research All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewResearchZosho AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/researchzosho?style=directory)](https://allmcps.com/mcp/researchzosho)
HTML Embed
<a href="https://allmcps.com/mcp/researchzosho"><img src="https://allmcps.com/api/badge/researchzosho?style=directory" alt="ResearchZosho on AllMCPs" /></a>