M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
O

o3 Search MCP Server

Community
MCP server that enables the use of OpenAI's high-end models and their powerful web search capabilities.
How to install ↓ Compare GitHub source repository ↗
Category
Other
Stars
★ 286 Very popular
Transport
stdio (local process)
Runtime
Node.js
Credentials
API key / credential required
License
MIT
Last commit
⚠
34FMRS · D

This server provides a convenient way to integrate OpenAI's advanced search capabilities via MCP, suitable for coding scenarios needing real-time info. However, it requires an OpenAI API key and possibly access permissions.

Strongest · Documentation 12/20 Weakest · Reliability 4/20

Reliability
4/20
Security and permissions
6/20
Maintenance
6/20
Documentation
12/20
Setup experience
6/20
Why each score
Reliability 4/20
README shows clear installation and configuration, but no source code or tests are available to verify actual behavior. No tests or CI configuration found, so core paths cannot be confirmed as reproducible.
Security and permissions 6/20
API key is passed via environment variables, not hardcoded. However, source code audit is missing, so data boundaries and potential leaks cannot be assessed. README does not mention least privilege or operation confirmation, but since it's a search proxy, risk is moderate.
Maintenance 6/20
Repository appears active (288 stars, 1 open issue), but no release history or dependency update records. MIT license present, but no security response channel found.
Documentation 12/20
README provides detailed installation, configuration, and example prompts, covering environment variables and endpoints. But troubleshooting and cost details are missing; documentation is above average but not comprehensive.
Setup experience 6/20
Clear setup examples for npx and local installation are provided. However, no runnable CI or tests, and setup requires manual steps, limiting the score.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the network

Best for

  • Developers using Claude Code or other AI coding agents to incorporate real-time information.
  • Developers needing to resolve rare errors or lacking documentation.
  • Developers needing design review or complex problem-solving support.

Not for

  • Users who don't want to depend on OpenAI API or lack an API key.
  • Users needing search capabilities via non-OpenAI models.
  • Users needing offline search or local search abilities.

Required permissions

  • Requires a valid OpenAI API key, and possibly API tier 4 or organization verification for o3 model access.
  • Network access to OpenAI API.
  • Write permissions for configuration directory during installation.

Risks and side effects

  • API key leakage: ensure key is kept secure and not committed to public repositories.
  • API usage may incur costs; monitor usage.
  • AI agent may provide inaccurate info; verify critical results.
  • Dependency on OpenAI service availability and potential rate limits.

Setup

Before you start

Runtime：Node.js

OPENAI_API_KEY requiredsecret Required OpenAI API key from the OpenAI platform; your account must have o3 access (Tier 4 or verified organization).
Other optional settings (5)
OPENAI_MODEL optional Model to use: o3, o4-mini, or gpt-5; defaults to o3.
SEARCH_CONTEXT_SIZE optional Search context size: low, medium, or high; defaults to medium.
REASONING_EFFORT optional Reasoning effort level: low, medium, or high; defaults to medium.
OPENAI_API_TIMEOUT optional API request timeout in milliseconds; defaults to 300000 (5 minutes).
OPENAI_MAX_RETRIES optional Max retries for failed requests; defaults to 3; SDK auto-retries on 429, 5xx, and connection errors.

Recommended installation via npx: run claude mcp add o3 -s user -e OPENAI_API_KEY=your-api-key -e OPENAI_MODEL=o3 -e SEARCH_CONTEXT_SIZE=medium -e REASONING_EFFORT=medium -e OPENAI_API_TIMEOUT=300000 -e OPENAI_MAX_RETRIES=3 -- npx o3-search-mcp. Alternatively, clone the repo and build locally: git clone [email protected]:yoshiko-pg/o3-search-mcp.git && cd o3-search-mcp && pnpm install && pnpm build, then configure to run node /path/to/o3-search-mcp/build/index.js.

claude_desktop_config.json 
{
  "mcpServers": {
    "o3-search": {
      "command": "npx",
      "args": [
        "o3-search-mcp"
      ],
      "env": {
        "OPENAI_API_KEY": "your-api-key",
        "OPENAI_MODEL": "o3",
        "SEARCH_CONTEXT_SIZE": "medium",
        "REASONING_EFFORT": "medium",
        "OPENAI_API_TIMEOUT": "300000",
        "OPENAI_MAX_RETRIES": "3"
      }
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "o3-search": {
      "command": "npx",
      "args": [
        "o3-search-mcp"
      ],
      "env": {
        "OPENAI_API_KEY": "your-api-key",
        "OPENAI_MODEL": "o3",
        "SEARCH_CONTEXT_SIZE": "medium",
        "REASONING_EFFORT": "medium",
        "OPENAI_API_TIMEOUT": "300000",
        "OPENAI_MAX_RETRIES": "3"
      }
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal
claude mcp add o3-search -e OPENAI_API_KEY=your-api-key -e OPENAI_MODEL=o3 -e SEARCH_CONTEXT_SIZE=medium -e REASONING_EFFORT=medium -e OPENAI_API_TIMEOUT=300000 -e OPENAI_MAX_RETRIES=3 -- npx o3-search-mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Register it with claude mcp add (or the JSON config), restart the client, and confirm the o3 search tool appears in the client's tool list; then ask a question the agent should delegate to o3 to verify end-to-end connectivity.

Troubleshooting

  1. If calls fail, check that OPENAI_API_KEY is correctly set and that your account has permission to access the selected model (e.g., o3).
  2. Verify network connectivity.
  3. Adjust OPENAI_MAX_RETRIES and OPENAI_API_TIMEOUT if encountering rate limits or timeouts.
  4. Ensure MCP configuration command and arguments are correct.

Things to try

Once connected, you can ask your AI assistant things like:

  • I'm getting this error on startup, please fix it. If it's too difficult, ask o3. (paste the error)
  • The WebSocket connection isn't working. Please debug it. If you don't know how, ask o3.
  • I want to upgrade this library to v2. Proceed while consulting with o3.
  • I want to build a collaborative editor, please design it and ask o3 for a design review.

Use cases

When stuck debugging, let o3 search broadly (e.g., GitHub issues, Stack Overflow) to find solutions.
Reference latest library information even without well-organized documentation.
Use as a sounding board for design reviews and complex task discussions.
AI agent autonomously decides to consult o3 when it deems necessary, expanding its problem-solving range.

Supported clients

Claude Desktop

Listed from the project's documentation, not tested by this site.

Overview

o3-search-mcp is an MCP server that allows AI coding agents (such as Claude Code) to autonomously consult OpenAI's advanced models for web search, helping solve complex problems. By integrating this server, agents can access real-time information from a wide range of sources, including GitHub issues and Stack Overflow, significantly increasing the chances of resolving niche issues. The server supports specifying OpenAI models (e.g., o3, o4-mini, gpt-5) and offers configurable parameters such as search context size and reasoning effort.

Similar servers

Telegram MCP Server 70 · B

Telegram MCP server powered by Telethon to let MCP clients read chats, manage groups, and send/modify messages, media, contacts, and settings.

★ 1.8k · Tools 76 Compare with this →
Compare with other MCP servers

Source revision 1ad6938fcfe5 Data synced 2026-10-12 Read the FMRS scoring method