# FetchPan

> FetchPan opens download links from file hosts (Rapidgator, Katfile, Nitroflare, Turbobit ...) and Chinese cloud drives (Baidu Netdisk, Quark, PikPak ...) and returns a direct download URL. A few files a day are free; faster downloads use paid traffic. FetchPan does not host files.

## For AI agents

- [API & MCP guide](https://fetchpan.com/developers): connect over MCP or REST. Step-by-step setup for Claude, ChatGPT, Claude Code, Codex and WorkBuddy.
- [Full developer docs as Markdown](https://fetchpan.com/llms-full.txt)
- [MCP server page](https://fetchpan.com/mcp-server): what it does, setup per client, tools. Also as [Markdown](https://fetchpan.com/mcp-server.md) (also served to requests that ask for text/markdown) and as a [JSON description](https://fetchpan.com/mcp-server.json).
- MCP server (streamable HTTP): https://fetchpan.com/mcp  It acts for the user's FetchPan account: apps with MCP sign-in (OAuth; Claude, ChatGPT, Claude Code, Cursor, VS Code) open https://fetchpan.com once to sign the user in. Claude Code: `claude mcp add --transport http fetchpan https://fetchpan.com/mcp`, then /mcp -> Authenticate.
- Or a free API token (sign in with Google at https://fetchpan.com/developers; header Authorization: Bearer fp_...). Either way: 3 free files a day, paid traffic for fast downloads.
- REST API base: https://fetchpan.com/v1/web  (same token). Flow: POST /jobs -> GET /jobs/{id} -> POST /jobs/{id}/download

## Pages

- [Supported sites and live status](https://fetchpan.com/sites)
- [Downloaders per site](https://fetchpan.com/downloaders)
- [Free vs Fast](https://fetchpan.com/pricing)
- [Questions](https://fetchpan.com/faq)
- [Copyright / abuse](https://fetchpan.com/copyright)

## MCP tools

- `download`: Give it a link; it waits for the file and returns a download URL. If the file is not ready in time, it returns a job id to continue with. A share with several files returns its list to pick one from.
- `check_link`: Is this site supported and working right now?
- `list_supported_sites`: All supported sites with live status.
- `list_files`: Open a folder of a Quark or PikPak share to find the file you want (then download it with share_id and file_id).
- `start_download`: Start a job without waiting. Returns a job id.
- `get_job`: Job status, and what a free or fast download would cost.
- `get_download_link`: The URL for a finished job (free or fast).
- `get_account`: Your email, traffic balance, free files per day and recent downloads.
- `list_packages`: Traffic packs you can buy.
- `create_payment_link`: A payment page for a pack, to give to you. Only a person can pay it.