Skip to main content
For AI agents: a documentation index is available at https://docs.parallel.ai/llms.txt. The full text of all docs is at https://docs.parallel.ai/llms-full.txt. You may also fetch any page as Markdown by appending .md to its URL or sending Accept: text/markdown.For Search and Extract, use /v1/search and /v1/extract for all new integrations and agent tool calls. Use /v1beta/search and /v1beta/extract only when maintaining an existing integration that already uses them. The request bodies changed too: on /v1/extract, full_content, fetch_policy and excerpt_settings go inside advanced_settings, and sending the v1beta top-level excerpts, full_content and fetch_policy to /v1/extract returns a 422. On /v1/search, search_queries is required (at least one short keyword query), and max_results, source_policy and fetch_policy go inside advanced_settings; sending only objective, or sending those fields at the top level, returns a 422. Do not substitute /v1 for the documented FindAll or Ingest endpoint paths.If you are evaluating Parallel Search or comparing it with other search APIs, read How to Eval Parallel Search before you build the harness, especially its Common pitfalls section. The most common mistake is sending the raw question as both objective and the only entry in search_queries. Write a goal-style objective and 1-3 short keyword queries instead.
The Parallel MCP server works in any MCP client that supports remote servers over Streamable HTTP. The server has two endpoints:
  • https://mcp.parallel.ai/mcp: the default endpoint. Without an API key, the tools the server lists include web_search, web_fetch, web_decisions, image_search (fast mode only), find_verticals, and search_vertical, at lower rate limits. An API key sent as a Bearer token (Authorization: Bearer <key>) adds web_research and raises rate limits.
  • https://mcp.parallel.ai/mcp?login: signs in with your Parallel account over OAuth when the client connects, and lists the default tools plus web_research. Use ?login to attribute every call to your Parallel account, for example in organization-wide deployments or Zero Data Retention (ZDR) setups.
Quote URLs that contain ? or & in shell commands. For authentication methods, tool selection, and override settings, see the quickstart. To call the server from your own code, see Programmatic use.

Claude Code

Run this command in your terminal:
In Claude Code, run /mcp to confirm the server is connected. To sign in with your Parallel account, add the OAuth endpoint instead:
Then run /mcp in Claude Code, select the server, and follow the steps in your browser to log in. For more details, see the Claude Code MCP documentation.

Codex CLI

Run one of the following, depending on how you want to authenticate:
Without an API key
With an API key
With OAuth
Each command writes the equivalent [mcp_servers.parallel] entry to ~/.codex/config.toml — you can also hand-edit it if you prefer. Restart Codex after adding the server. For more details, see the Codex MCP documentation.

Claude Desktop / Claude.ai

The easiest path is the Parallel Search listing in Claude’s connector directory. Open the listing and click Add to Claude, or search for “Parallel Search” under Settings → Connectors → Browse connectors. The steps below add the server as a custom connector instead.
  • Pro / Max plans: go to Customize → Connectors → ”+ Add” → “Add custom connector”.
  • Team / Enterprise plans: an Owner or Primary Owner must first go to Organization settings → Connectors → “Add” → “Custom” → “Web” to add the connector for the organization.
Enter the following connector details and complete the setup prompts:
To sign in with your Parallel account and use web_research, use https://mcp.parallel.ai/mcp?login instead. Team / Enterprise members: after your owner adds the connector, go to Customize → Connectors, find it in the list, and click “Connect” to authenticate. Contact your organization owner if the connector is not available.
Org admins (Team / Enterprise plans): to route Claude’s web queries through the Parallel MCP instead of Claude’s built-in web search, disable the native web search tool at Organization settings → Capabilities. Once off at the workspace level, members cannot re-enable it individually. See the Claude web search docs.For the org-wide deployment, also point the connector at the /mcp?login URL (not /mcp) so every request requires a Parallel API key or OAuth — useful when you need every call attributed to the organization.
If you are not an admin, in Claude Desktop go to Settings → Developer → Edit Config and use the following JSON. If your config already has an mcpServers object, add the Parallel entry inside it without replacing your other servers.
No API key is required for web_search and web_fetch. To use web_research and raise your rate limits, grab a key from Platform and append "--header", "authorization: Bearer YOUR-PARALLEL-API-KEY" to the args array. For more details, see the Claude remote MCP documentation.

Cursor

Add to ~/.cursor/mcp.json or .cursor/mcp.json (project-specific):
For more details, see the Cursor MCP documentation.

VS Code

The quickest path is the guided flow — no file editing required. Run MCP: Add Server from the Command Palette (⇧⌘P / Ctrl+Shift+P), pick HTTP, and paste the server URL:
VS Code then prompts for a server name (Parallel) and whether to add it as Global (available in every workspace) or Workspace (this project only), and writes the configuration for you. Confirm the trust prompt the first time the server starts. To configure it by hand instead, add the following to .vscode/mcp.json in your workspace, or to the user-level mcp.json that the MCP: Open User Configuration command opens:
The user-level mcp.json lives inside your VS Code profile folder, not at ~/.vscode/mcp.json. Open it with MCP: Open User Configuration rather than typing the path — a file you create at ~/.vscode/mcp.json is ignored, with no error to tell you why.
For more details, see the VS Code MCP documentation.

fx (Vercel)

fx connects to the Parallel MCP through mcp-remote. This setup requires Node.js and npx. Run this command inside fx:
fx saves the server to ~/.fx/mcp.json without removing your existing MCP servers and reloads the configuration. Run /mcp list to confirm parallel is listed and check its status. To use web_research and raise your rate limits, get an API key from platform.parallel.ai and export PARALLEL_API_KEY before launching fx. Add "--header" and "Authorization: Bearer ${PARALLEL_API_KEY}" to the parallel entry’s args array in ~/.fx/mcp.json, then run /mcp list. fx loads a changed mcp.json on your next prompt or /mcp command. For more details, see the fx MCP documentation.

Other Clients

Add to ~/.codeium/windsurf/mcp_config.json:
For more details, see the Windsurf MCP documentation.
In the Cline panel, open MCP Servers → Configure → Configure MCP Servers, then add the following under mcpServers:
For more details, see the Cline MCP documentation.
Add to ~/.gemini/settings.json:
For more details, see the Gemini CLI MCP documentation.
The easiest path is the Parallel Search app in the ChatGPT app directory. Open the listing, or search for “Parallel Search” at chatgpt.com/apps, and click Install. The steps below add the server as a custom connector instead.Warning: Developer Mode must be enabled, and this feature may not be available to everyone. MCPs in ChatGPT are experimental and may not work reliably.First, go to Settings → Connectors → Advanced Settings, and turn on Developer Mode.Then, in connector settings, click Create and fill in one of the following. For use without an account, pair the default URL with No authentication:
To use every tool, including web_research, sign in with OAuth:
In a new chat, ensure Developer Mode is turned on with the connector(s) selected.For more details, see the ChatGPT Developer Mode documentation.
Run this command in your terminal:
web_search and web_fetch work without signing in. To sign in with OAuth, add the OAuth endpoint instead:
Amp starts the OAuth flow in your browser the next time you start it.For more details, see the Amp MCP documentation.
Add to .kiro/settings/mcp.json (workspace) or ~/.kiro/settings/mcp.json (global):
For more details, see the Kiro MCP documentation.
In the Antigravity Agent pane, click the menu (⋮) → MCP Servers → Manage MCP Servers → View raw config, then add:
To use web_research and raise your rate limits, add a headers block with "Authorization": "Bearer YOUR_API_KEY" and your key from platform.parallel.ai.For more details, see the Antigravity MCP documentation.
Add to opencode.json (project) or ~/.config/opencode/opencode.json (global):
For more details, see the OpenCode MCP documentation.
Add to .roo/mcp.json in your workspace (or edit the global mcp_settings.json from the Roo Code MCP settings view):
Project config takes precedence over global. For more details, see the Roo Code MCP documentation.
Run this command in your terminal:
Or add manually to ~/.openhands/mcp.json:
Inside an OpenHands conversation, use /mcp to verify the server is active. For more details, see the OpenHands MCP documentation.
Run this command in your terminal:
Inside droid, type /mcp to open the interactive manager and verify the server is connected. For more details, see the Factory MCP documentation.
Pi has built-in MCP support. Add the server in your terminal:
If Pi is already open, run /reload in that session after adding the server. Then run /mcp inside Pi to verify the connection. See the Pi MCP documentation.Alternatively, install the pi-mcp-adapter package:
Restart Pi, then add the Parallel MCP to .mcp.json in your project (or ~/.config/mcp/mcp.json for a user-global config):
directTools: true registers the Parallel tools alongside Pi’s built-in tools instead of hiding them behind the adapter’s mcp proxy. Run /mcp inside Pi to verify the server is detected. For more details, see the pi-mcp-adapter README.Alternatively, skip MCP entirely and use the Parallel CLI as a Pi skill — that’s the same pattern as our ClawHub / OpenClaw integration.
OpenClaw stores MCP server definitions in ~/.openclaw/openclaw.json under mcp.servers. Save the Parallel MCP via the openclaw mcp set CLI:
Or edit ~/.openclaw/openclaw.json directly:
The transport field is required — OpenClaw defaults to SSE when it’s omitted, but the Parallel MCP uses Streamable HTTP. No API key is required for web_search and web_fetch; to use web_research and raise your rate limits, add an optional headers map with "Authorization": "Bearer YOUR-PARALLEL-API-KEY" (key from platform.parallel.ai).openclaw mcp set only writes to config — it doesn’t connect to the server or reload running agents. Start a new OpenClaw agent session (or restart your current one) for the tools to show up. Verify the saved definition with openclaw mcp list / openclaw mcp show parallel. For the full MCP CLI reference, see the OpenClaw MCP docs.You can also skip MCP entirely and use the Parallel CLI as an OpenClaw skill — see the ClawHub integration.
Add to ~/.hermes/config.yaml:
For more details, see the Hermes Agent MCP documentation.
Add a file at .continue/mcpServers/parallel.yaml in your workspace:
For more details, see the Continue.dev MCP documentation.
Open Settings → AI → MCP Servers, click Add Server, and choose Add Remote Server. Or add the entry to your Zed settings.json (open it with the zed: open settings file action). Zed uses a top-level context_servers object:
To use web_research and raise your rate limits, add a headers object with "Authorization": "Bearer YOUR-PARALLEL-API-KEY" and a key from platform.parallel.ai. To sign in with OAuth instead, set url to https://mcp.parallel.ai/mcp?login and leave out the Authorization header. Zed then prompts you to authenticate in your browser.For more details, see the Zed MCP documentation.
Open Settings → Agents → MCP servers, click + Add, and paste:
You can also save the same entry to ~/.warp/.mcp.json (global) or .warp/.mcp.json at the project root. To use web_research and raise your rate limits, add a headers object with "Authorization": "Bearer YOUR-PARALLEL-API-KEY" and a key from platform.parallel.ai. To sign in with OAuth instead, set url to https://mcp.parallel.ai/mcp?login. Warp opens a browser sign-in flow when the server starts.For more details, see the Warp MCP documentation.
Run the Install MCP Server command and fill in:
To use web_research and raise your rate limits, add an HTTP header Authorization: Bearer YOUR-PARALLEL-API-KEY with a key from platform.parallel.ai. To sign in with OAuth instead, set the URL to https://mcp.parallel.ai/mcp?login and keep the OAuth type on Dynamic. Raycast then prompts you to Sign In before the tools become available.For more details, see the Raycast MCP documentation.
Add to LM Studio installs the server in one click.
Add to Goose installs the server in one click.
If your client isn’t listed, most MCP clients accept the mcpServers format shown in the Cursor section. For clients that use a different wrapper (e.g., VS Code’s top-level servers, Windsurf’s serverUrl), check the client’s documentation for the correct field names.

Troubleshooting

Common Installation Issues

This error occurs when Cline attempts OAuth authentication but the redirect URI isn’t using HTTPS. web_search and web_fetch at /mcp don’t require OAuth — wrap it in mcp-remote so Cline treats it as a plain stdio server:
If you want web_research or higher rate limits, append "--header", "authorization: Bearer YOUR-PARALLEL-API-KEY" to the args array with a key from platform.parallel.ai.
Gemini CLI uses HTTP MCPs and can authenticate via OAuth. web_search and web_fetch at /mcp don’t require an API key or OAuth, so the simplest setup has no auth at all:
Add this to ~/.gemini/settings.json. If you want web_research or higher rate limits, add a headers object and pass a key from platform.parallel.ai as a Bearer header:
VS Code reads two mcp.json locations: .vscode/mcp.json inside the workspace, and a user-level mcp.json in your VS Code profile folder, which the MCP: Open User Configuration command opens. ~/.vscode/mcp.json is not one of them — that directory holds argv.json and your installed extensions, so a server defined there never loads and VS Code reports no error.If a hand-written config isn’t picked up, run MCP: Open User Configuration and check whether the file that opens is the one you edited. Run MCP: List Servers to confirm VS Code sees the server, and use MCP: Add Server to let VS Code write the entry to the correct file.
VS Code’s mcp.json uses a different structure than Cursor’s mcp.json. Common mistake: copying a Cursor-style config into VS Code.Incorrect (Cursor format):
Correct (VS Code format, .vscode/mcp.json or user-level mcp.json):
Note: VS Code uses a top-level servers object (not mcpServers) and includes type: "http" for remote HTTP servers.
Windsurf uses a different configuration format than Cursor.Correct Windsurf configuration:
Note: Windsurf uses serverUrl instead of url. Add this to your Windsurf MCP configuration file.
If you’re getting connection errors:
  1. Check your network: Ensure you can reach https://mcp.parallel.ai
  2. Verify API key: Make sure your key is valid at platform.parallel.ai
  3. Check balance: A 402 error means insufficient credits—add funds to your account
  4. Restart your IDE: Some clients cache MCP connections
If the MCP installs but tools don’t show up:
  1. Restart your IDE completely (not just reload)
  2. Check configuration syntax: Ensure valid JSON with no trailing commas
  3. Verify the server URL: The base must be https://mcp.parallel.ai/mcp (no trailing slash), with documented query parameters such as ?login allowed
  4. Check IDE logs: Look for MCP-related errors in your IDE’s output/debug panel