Parallel Search MCP: free web search for an agent
Parallel Search MCP
Parallel's official remote MCP server: an agent searches the web and reads pages with no account or key, free tier rate limited
- 33 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- Query text goes to the third-party Parallel service, so confidential data is best kept out of search
- Page content from the open web lands in the agent context and may carry instructions from third parties
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/parallel-search-mcpDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- Query text goes to the third-party Parallel service, so confidential data is best kept out of search
- Page content from the open web lands in the agent context and may carry instructions from third parties
This is third-party code. Review the repository files before installing.
What it does
The server gives an agent web search and page reading through Parallel, a search service for AI agents. It is a remote server at https://search.parallel.ai/mcp over streamable HTTP, with no account, API key or sign-in. It has two tools. The first searches the web and returns relevant excerpts with source links. The second reads given addresses and returns page content. One search is often enough, and the agent opens a page when it needs detail. The free tier is meant for personal agents, hobby projects and trials and is rate limited. For higher limits and production workloads you create an account at platform.parallel.ai and connect the Parallel Search connector with sign-in. For Gemini CLI there is an extension, and Hermes and OpenClaw have separate packages.
Who it is for. Anyone using coding agents who wants to give them current search over documentation and facts without sign-up or key setup.
Good fit when
- An agent writes code for a library and needs current official documentation with a link
- You want to check the latest Node.js version and read the release notes
- An agent has to read one specific page and retell it on point
- A hobby project needs search for an agent without accounts or keys
- You try web search in Codex or Cursor before choosing a paid service
Not a fit when
- You need search over closed sources and internal databases: the server searches the open web
- A production workload with heavy request volume: the free tier is rate limited
- Queries contain confidential data: they go to a third-party service
- You need an availability guarantee under contract: the free tier is meant for trials
Example request
Find the official FastAPI documentation on response models, show a short example and link to the source.MCP
- Transport
- http
- Authentication
- not required
Limitations
The free tier is anonymous and rate limited, and the README does not state exact limits. The repository holds instructions and client configs and does not include server code: search runs on Parallel's side. The README asks you not to add an API key or point the plugin at an authenticated endpoint, since a separate connector serves that.
Using the Hermes package together with a manual connection duplicates the tools, so pick one.
How to disable. Remove the connection with claude mcp remove parallel-search or delete the parallel-search entry from the client config. For Gemini CLI remove the extension, for Hermes run hermes plugins remove parallel-search.
FAQ
How do I connect Parallel Search MCP to Codex?
With codex mcp add parallel-search --url https://search.parallel.ai/mcp. Then start a new session and run /mcp to check the connection. No account or key is needed.
Do I need an account or an API key?
No, both tools work anonymously within the free limits. An account at platform.parallel.ai is needed only for higher limits, analytics and production workloads, and then a separate connector is connected.
Which tools does it have?
Two: web_search searches the web and returns excerpts with links, and web_fetch reads given addresses and returns page content. One search is often enough, and a page is opened when needed.
Which clients does it work with?
With clients that support remote Streamable HTTP servers: Cursor, VS Code, Codex, OpenCode and others. There is an extension for Gemini CLI and separate packages for Hermes and OpenClaw from those projects' catalogs.
What is the free tier for?
Personal agents, hobby projects and getting to know the service. Requests are rate limited, and for production workloads Parallel suggests an account with higher limits and tariff billing.
Discussion
No questions or reviews yet.
Related
MCP serversThe Parallel plugin for Claude Code and Codex: web search and page fetching via MCP, plus deep research and monitoring skills.
Metasearch for agents: one MCP server and HTTP API merge results from many search providers into unified JSON with deduplication.
A unified search MCP server: one interface to Tavily, Brave, Kagi, Exa, Linkup, Firecrawl and GitHub search
MCP server for real-time web search: general and vertical search, batch queries and page content extraction
A local MCP server with no API keys: multi-engine web search, page fetching and document reading tuned for LLMs.
Linkup MCP server: real-time web search, page fetching and long research tasks with sources for an agent
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Parallel Search MCP: free web search for an agent (this card) | yes | not required | 33 | |
| Exa MCP Server | yes | OAuth | 5,231 | Exa search with an API key and extended tools, here two tools without a key |
| Brave Search MCP Server | yes | API key | 1,577 | Brave search by API key, while this server is open without sign-up |
| Parallel Agent Skills | no | 93 | Parallel skills for agents, while here a remote MCP search server |
README in short
The README briefly describes free web search, gives a ready instruction for an agent to install it and links for Cursor and VS Code. Then come manual setups for Codex, Gemini CLI, Hermes, OpenCode and OpenClaw, example requests, a table of two tools and a section on free access.