Step-by-step setup guides
The picker below gives you the snippet. These pages walk the whole thing: where the config file lives, how to confirm the tools are live, and what each error means.
- How to add web search to Claude Code
- How to add web search to Claude Desktop
- How to add web search to Cursor
- How to add web search to LM Studio
- How to add web search to OpenCode
- How to add web search to Codex CLI
- How to add web search to Gemini CLI
- How to add web search to Cline
- How to add web search to Windsurf
- How to add web search to Continue
- How to add web search to Zed
- How to add web search to Goose
- How to add web search to OpenClaw
- How to add web search to ChatGPT
- How to add web search to Ollama
- How to add web search to LangChain
- How to add web search to LlamaIndex
- How to add web search to n8n
Integrations
Connect any agent in under a minute. One key, one endpoint — pick your tool for a copy-paste snippet, plus a link to its official docs.
Almost every tool below is MCP-native: point it at https://mcp.blopus.ai/mcp/ with an Authorization: Bearer header and the search + fetch tools appear automatically. The generic config is:
Agents & code editors#
Register the hosted MCP server from the Claude Code CLI — search + fetch land instantly.
Drop Blopus into ~/.cursor/mcp.json (global) or .cursor/mcp.json in a project.
Add Blopus in Cline's MCP settings — the Blopus tools appear in every task.
Add Blopus as an MCP server in your Continue config and it's available to every model.
Register Blopus as a context server in Zed's settings and the agent can search the live web.
Chat apps & desktop#
Settings → Connectors → Add custom connector, then paste the URL and your key.
Turn on Developer mode and add Blopus as a connector, or call it as a function from the API.
LM Studio is an MCP host — open Program → Edit mcp.json and add Blopus.
Jan is an MCP host — turn on MCP Servers in Settings and add Blopus for web search + fetch.
Add Blopus under mcpServers in librechat.yaml and it's available to every agent.
Open WebUI is an MCP client (v0.6.31+) — add Blopus as a Streamable-HTTP tool server with a Bearer token.
Frameworks & SDKs#
Load the Blopus tool spec and hand it to any LlamaIndex agent.
Load Blopus via langchain-mcp-adapters (or a plain @tool) and pass it to create_react_agent.
Load Blopus with mcp_server_tools over Streamable HTTP and hand the tools to an AssistantAgent.
No-code & automation#
Wire Blopus into any workflow with the MCP Client Tool node — no code.
Drop a Custom MCP node on the canvas, point it at the Blopus endpoint, and refresh available actions.
Add the MCP Tools component (Streamable HTTP) and the agent gets Blopus search + fetch.
Create a Custom tool from Blopus' OpenAPI schema, then enable it in any agent or workflow.
LLM serving engines#
Any OpenAI-compatible server can call Blopus through function / tool calling. Register a search tool whose handler hits the Blopus REST API, and your local model reaches the live web:
Works the same across these OpenAI-compatible engines:
Don't see your tool? If it speaks MCP or can call a function, Blopus drops in — same key, one endpoint, no lock-in. See the MCP server reference for the full contract.