Skip to content
Back to search

deep-research-mcp

This repo is an experiment on agent coding. 95% of the code is written by LLM's

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/teelaitila--deep-research-mcp/badge.svg)](/listings/teelaitila--deep-research-mcp)

HTML

<a href="/listings/teelaitila--deep-research-mcp"><img src="/api/listings/teelaitila--deep-research-mcp/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

1. Clone and install:

git clone https://github.com/Ozamatash/deep-research
cd deep-research
npm install

2. Set up environment in `.env.local`:

# Copy the example environment file
cp .env.example .env.local

3. Build:

# Build the server
npm run build

4. Run the cli version:

npm run start

5. Test MCP Server with Claude Desktop: Follow the guide thats at the bottom of server quickstart to add the server to Claude Desktop: https://modelcontextprotocol.io/quickstart/server

For remote servers: Streamable HTTP

npm run start:http

Server runs on `http://localhost:3000/mcp` without session management.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 178 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # DISCLAIMER
  2. This repo is an experiment on agent coding. 95% of the code is written by LLM's
  3. # Open Deep Research MCP Server
  4. An AI-powered research assistant that performs deep, iterative research on any topic. It combines search engines, web scraping, and AI to explore topics in depth and generate comprehensive reports. Available as a Model Context Protocol (MCP) tool or standalone CLI. Look at exampleout.md to see what a report might look like.
  5. ## Quick Start
  6. 1. Clone and install:
  7. ```bash
  8. git clone https://github.com/Ozamatash/deep-research
  9. cd deep-research
  10. npm install
  11. ```
  12. 2. Set up environment in `.env.local`:
  13. ```bash
  14. # Copy the example environment file
  15. cp .env.example .env.local
  16. ```
  17. 3. Build:
  18. ```bash
  19. # Build the server
  20. npm run build
  21. ```
  22. 4. Run the cli version:
  23. ```bash
  24. npm run start
  25. ```
  26. 5. Test MCP Server with Claude Desktop:
  27. Follow the guide thats at the bottom of server quickstart to add the server to Claude Desktop:
  28. https://modelcontextprotocol.io/quickstart/server
  29. For remote servers: Streamable HTTP
  30. ```bash
  31. npm run start:http
  32. ```
  33. Server runs on `http://localhost:3000/mcp` without session management.
  34. ## Features
  35. - Performs deep, iterative research by generating targeted search queries
  36. - Controls research scope with depth (how deep) and breadth (how wide) parameters
  37. - Evaluates source reliability with detailed scoring (0-1) and reasoning
  38. - Prioritizes high-reliability sources (≥0.7) and verifies less reliable information
  39. - Generates follow-up questions to better understand research needs
  40. - Produces detailed markdown reports with findings, sources, and reliability assessments
  41. - Available as a Model Context Protocol (MCP) tool for AI agents
  42. - For now MCP version doesn't ask follow up questions
  43. - Natural-language source preferences (avoid listicles, forums, affiliate reviews, specific domains)
  44. ### Model Selection (OpenAI, Anthropic, Google, xAI)
  45. Pick a provider and model per run.
  46. - CLI: you will be prompted for provider and model. Example: `openai` + `gpt-5.2`.
  47. - MCP/HTTP: pass `model`, e.g. `openai:gpt-5.2` (also accepts `openai/gpt-5.2`).
  48. Set the corresponding API key in `.env.local`:
  49. ```
  50. OPENAI_API_KEY=...
  51. ANTHROPIC_API_KEY=...
  52. GOOGLE_API_KEY=...
  53. XAI_API_KEY=...
  54. ```
  55. Optionally set default models per provider:
  56. ```
  57. OPENAI_MODEL=gpt-5.2
  58. ANTHROPIC_MODEL=claude-opus-4-5
  59. GOOGLE_MODEL=gemini-3-pro-preview
  60. XAI_MODEL=grok-4-1-fast-reasoning
  61. ```
  62. If you use a non-default OpenAI endpoint:
  63. ```
  64. OPENAI_ENDPOINT=https://api.openai.com/v1
  65. ```
  66. ## How It Works
  67. ```mermaid
  68. flowchart TB
  69. subgraph Input
  70. Q[User Query]
  71. B[Breadth Parameter]
  72. D[Depth Parameter]
  73. FQ[Feedback Questions]
  74. end
  75. subgraph Research[Deep Research]
  76. direction TB
  77. SQ[Generate SERP Queries]
  78. SR[Search]
  79. RE[Source Reliability Evaluation]
  80. PR[Process Results]
  81. end
  82. subgraph Results[Research Output]
  83. direction TB
  84. L((Learnings with
  85. Reliability Scores))
  86. SM((Source Metadata))
  87. ND((Next Directions:
  88. Prior Goals,
  89. New Questions))
  90. end
  91. %% Main Flow
  92. Q & FQ --> CQ[Combined Query]
  93. CQ & B & D --> SQ
  94. SQ --> SR
  95. SR --> RE
  96. RE --> PR
  97. %% Results Flow
  98. PR --> L
  99. PR --> SM
  100. PR --> ND
  101. %% Depth Decision and Recursion
  102. L & ND --> DP{depth > 0?}
  103. DP -->|Yes| SQ
  104. %% Final Output
  105. DP -->|No| MR[Markdown Report]
  106. %% Styling
  107. classDef input fill:#7bed9f,stroke:#2ed573,color:black
  108. classDef process fill:#70a1ff,stroke:#1e90ff,color:black
  109. classDef output fill:#ff4757,stroke:#ff6b81,color:black
  110. classDef results fill:#a8e6cf,stroke:#3b7a57,color:black,width:150px,height:150px
  111. class Q,B,D,FQ input
  112. class SQ,SR,RE,PR process
  113. class MR output
  114. class L,SM,ND results
  115. ```
  116. ## Advanced Setup
  117. ### Using Local Firecrawl (Free Option)
  118. Instead of using the Firecrawl API, you can run a local instance. You can use the official repo or my fork which uses searXNG as the search backend to avoid using a searchapi key:
  119. 1. Set up local Firecrawl:
  120. ```bash
  121. git clone https://github.com/Ozamatash/localfirecrawl
  122. cd localfirecrawl
  123. # Follow setup in localfirecrawl README
  124. ```
  125. 2. Update `.env.local`:
  126. ```bash
  127. FIRECRAWL_BASE_URL="http://localhost:3002"
  128. ```
  129. ### Optional: Observability
  130. Add observability to track research flows, queries, and results using Langfuse:
  131. ```bash
  132. # Add to .env.local
  133. LANGFUSE_PUBLIC_KEY="your_langfuse_public_key"
  134. LANGFUSE_SECRET_KEY="your_langfuse_secret_key"
  135. ```
  136. The app works normally without observability if no Langfuse keys are provided.
  137. ## License
  138. MIT License

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.