TinyFish MCP server
Local proxy to the hosted TinyFish web agent MCP
Official TinyFish local MCP proxy (@tiny-fish/mcp): loopback Streamable-HTTP at 127.0.0.1:3711 that relays to the hosted web agent at agent.tinyfish.ai/mcp — search, fetch, and browser automation.
TinyFish (@tiny-fish/mcp) is the official local MCP server: a transparent reverse proxy that exposes Streamable-HTTP at http://127.0.0.1:3711/mcp and forwards every request to https://agent.tinyfish.ai/mcp. The proxy defines no schemas of its own — tools/list, tools/call, resources, errors, and the run_web_automation SSE progress stream all come from the hosted server. Prefer the hosted URL when your client supports remote MCP; use this package when you need a local endpoint or TINYFISH_API_KEY auth for CI/headless. Upstream tools include free search and fetch_content plus wallet-metered web automation and browser sessions.
Transparent reverse proxy
Streaming (SSE) and JSON responses are relayed content-identical to upstream — whatever the hosted server exposes is what your client sees.
Hosted tools, free search/fetch
Relays search, fetch_content, run_web_automation, browser sessions, and wallet tools documented at docs.tinyfish.ai/mcp-integration.
API-key auth for headless use
Reads TINYFISH_API_KEY and sends X-API-Key upstream — for CI and scripts instead of hosted OAuth.
Loopback-only trust boundary
Binds 127.0.0.1 and rejects non-localhost Origin headers (403), mitigating DNS-rebinding.
Who it’s for. Teams whose MCP client only talks to local servers, or who need TinyFish API-key auth from the environment (CI, headless) rather than hosted OAuth — plus anyone connecting agents to live-web search, fetch, and browser automation.
Example prompts
auth: TINYFISH_API_KEY (proxy) or OAuth (hosted) · pricing: Search/Fetch free · Agent $0.016/step · Browser $0.002/min · reviewed by MCPExplorer
How to read this: tool names here are observed from a live tools/list handshake. The Risk label is a heuristic inferred from the tool name (write/destructive verbs), not from executing the tool — a conservative guess, not a verified capability. We never escalate risk from a description. Found one that's wrong? Tell us — we fix on report.
| Tool | Risk | Side effects | Approval |
|---|---|---|---|
| batch_cancel Cancel multiple running or pending runs by their IDs (up to 8). Idempotent. | destructive | true | true |
| batch_status Check the status of multiple runs at once (up to 8). Poll every 30–60 seconds until all_terminal is true. | write | true | unknown |
| cancel_run Cancel a running or pending automation run by ID. Idempotent — already-terminal runs return their current status. | destructive | true | true |
| close_browser_session Close a browser session by ID to release it early. Idempotent. | destructive | true | true |
| create_browser_session Create a remote stealth Chrome browser session and get CDP connection details for Playwright/Puppeteer/Selenium. | write | true | unknown |
| fetch_content Fetch web pages, render JS-heavy pages when needed, and extract clean structured content (markdown/html/json). Up to 10 URLs per request. Free tier. Preferred over automation when you only need to read page content. | write | true | unknown |
| get_run Retrieve details of a specific automation run by ID. Returns status, result, and error info. | write | true | unknown |
| get_search_usage List past search usage records with filtering by date range and status. | write | true | unknown |
| get_wallet Read-only. Returns current wallet balance, auto-reload state, per-product contract rates, and any in-flight top-up. | read | false | unknown |
| list_browser_sessions List your browser sessions with optional filtering by session ID, time range, and status. | read | false | unknown |
| list_fetch_usage List past fetch requests with filtering by date range and status. Returns metadata only. | write | true | unknown |
| list_runs List your automation runs with optional filtering by status and pagination. | read | false | unknown |
| run_web_automation Execute multi-step web automation given a URL and natural language goal. Streams progress in real-time. Use for clicking, navigating, filling forms, logging in, or interacting with a site. Pass use_profile/profile_id for saved Browser Context Profiles. | write | true | unknown |
| run_web_automation_async Same as run_web_automation but returns a run_id immediately without waiting. Poll with get_run. Supports use_profile, profile_id, and use_vault. | write | true | unknown |
| search Search the web and get structured results with titles, snippets, and URLs. Free tier. Optional filters: recency, dates, domain_type, location, language, include/exclude domains, page, purpose. | write | true | unknown |
- repohttps://github.com/tinyfish-io/tinyfish-mcp-server
- packagehttps://www.npmjs.com/package/@tiny-fish/mcp
- homepagehttps://tinyfish.ai
- licenseMIT
- adoption2 stars · 2 forks
The access this server can exercise, inferred from its verified tools — not a declared OAuth scope.
Add the “as seen on MCPExplorer” badge to your README. 
This is one server. A loadout combines the right servers, governance, and proven plays for a whole job — assembled deliberately, not tool-dumped.
Explore loadouts →