One media library for all your AI

Upload once.

Find, use and share across your AI workflows.

drop. is the one place your images, PDFs and files live, for you and every agent you work with. Add a file from your browser, your terminal, Claude, ChatGPT or Cursor, and any of them can find it, use it and hand out a link.

Start dropping → Connect an agent
Free to start · cancel any time ·
q3-revenue.png
Drop images or PDFs here
https://cdn.drop.media/8f2c41/q3-revenue.png
PNG · 214 KB public CDN
Put the Q3 revenue chart in the board update.
drop · list_files  “q3 revenue” ✓ 1 file
Found it — the chart you uploaded on Tuesday. It’s in the update: ![Q3 revenue](https://cdn.drop.media/8f2c41/q3-revenue.png)
$ drop ls board
9b11f0 2026-10-22 (use: drop link 9b11f0) [private]
$ drop link 9b11f0 --ttl 1d
https://drop.media/p/9b11f0?signature=…
# the deck Claude saved · signed link for a day
npm i -g drop-media no dependencies

Hosting files for teams at

BrewDog Silentnight eKitchens devsbydevs
How it works

One library. Four ways in.

Whoever adds a file — you, a script or an agent — it lands in the same library. Everyone else can find it there and get a URL that works in a PR, a doc, a chat or a CMS.

01 / Web

Drag, drop, find.

Drop files on the page and copy the link. Everything your agents added is under Media too, ready to search, rename and tidy.

Drop files hereImages, PDF, video, audio, text, Office
02 / MCP

Plug in your agent.

A remote MCP server that lets Claude, ChatGPT, Cursor and friends search, upload, share and generate.

$ claude mcp add \ --transport http drop \ https://drop.media/mcp \ --header "Authorization: Bearer $KEY"
03 / CLI

Pipe it from anywhere.

One command prints a link. Add -m for ready-made markdown.

$ drop *.png -m # ![chart](https://cdn.drop…) $ pbpaste | drop - \ --filename clip.png
04 / API

Build on it.

A small REST API with scoped keys. Multipart, base64 or fetch-from-URL.

$ curl drop.media/api/v1/files \ -H "Authorization: Bearer $KEY" \ -F file=@chart.png \ -F expires_in=7d
Built for agents

Give all your agents the same library.

Agents make screenshots, charts and reports all day, then lose them when the chat ends. With drop. they land in one library that every agent you connect can search, reuse and link to — without ever touching your cloud credentials.

  1. 1

    Create a scoped key

    Pick exactly what it can do: read, write, delete, generate. Revoke it in one click.

  2. 2

    Add the MCP server

    One command for Claude Code, one config block for everything else.

  3. 3

    Ask for your files

    “Use the logo I uploaded yesterday.” “Send me that report, private, for a day.” Done.

ClaudeChatGPTCursorAny MCP clientShell-capable agents, via the CLI
drop.media/mcp · 7 tools connected
upload_filefiles:write

Host a local file, base64 blob or remote URL. Public or private, with optional expiry.

create_upload_linkfiles:write

A one-time URL for curl -T — big files never pass through the model.

generate_imageimages:generate

Create an image from a prompt and host it. The prompt is saved alongside.

share_filefiles:read

Get a fresh signed link to a private file, valid from minutes up to 7 days.

list_files get_filefiles:read

Find what was uploaded before, by name or type.

delete_filefiles:delete

Clean up after itself.

claude mcp add --transport http drop https://drop.media/mcp --header "Authorization: Bearer <key>"
Or paste this into any AI 
drop.media is my media library, shared across all my AI tools: upload a file once, then find, use and share it from any of them. Use it to host files and get a shareable link (screenshots for pull requests and issues, screen recordings, charts and reports, CSVs, images for docs, chat and web pages), and to find files I or another agent added before. Use whichever of these you have:

1. MCP (best). If you have the drop.media MCP server (https://drop.media/mcp), use its tools: create_upload_link for a file on disk (then run the curl command it returns), upload_file for a URL or base64 content, generate_image to create an image, and list_files, get_file, share_file and delete_file to manage them. If it isn't set up, ask me to add https://drop.media/mcp as a connector or app (in ChatGPT or Claude, for example) and sign in once; it uses OAuth, so it keeps working in later conversations. Clients without OAuth can use an API key from https://drop.media/developers instead, e.g. for Claude Code:
   claude mcp add --transport http drop https://drop.media/mcp --header "Authorization: Bearer <api key>"

2. CLI. If you can run shell commands: npm install -g drop-media, then drop login <api key>. After that, drop screenshot.png prints the URL (add -m for a markdown snippet) and drop gen "<prompt>" generates an image.

3. REST API. POST https://drop.media/api/v1/uploads with the header "Authorization: Bearer <api key>" and the JSON body {"filename": "screenshot.png"} returns an upload_url. Then run curl --fail-with-body -T ./screenshot.png '<upload_url>', which prints JSON with the hosted file's url. If it times out, run it again: the same file isn't stored twice.

4. In my browser. If you're working in my browser on drop.media, the pages offer WebMCP tools that act as me, including create_upload_link, upload_file, list_files and generate_image.

Guidelines:
- When I mention a file I already have (a logo, a chart, last week's report), look for it with list_files before making a new one.
- Files are public by default, with a permanent CDN URL. Use visibility "private" for anything sensitive and share it with share_file, which gives an expiring link.
- Set expires_in (e.g. "7d") on throwaway files so they're deleted automatically.
- Prefer upload links to base64 so files arrive byte for byte.
- Give me back the URL, or the markdown snippet from the response.

Full reference: https://drop.media/llms.txt
What you get

One library. Sharp edges filed off.

Find it from anywhere

Search by name, type or the prompt that made it — on the web, in the terminal, over the API or from any connected agent.

One library, every agent

Private when it matters

Private files never touch the public CDN. Share them with signed links that expire.

1 hour → 7 days

Files that tidy themselves

Set an expiry when you upload and the file deletes itself. Perfect for throwaway screenshots.

30m · 1h · 7d · up to 365d

Image generation

Describe an image, get a hosted one. The prompt and model are kept with the file.

50 a day included

Keys with limits

Every API key carries only the abilities you give it. Agents see only the tools their key allows.

Fast links that stay put

Public files are served from cdn.drop.media until you remove them. Swap the file behind a link without changing it.

Free
$0

Your library, shared by you and your agents.

  • Files up to 10 MB
  • Your media library, on the web and in every agent
  • 50 uploads in total
  • 500 MB of storage
  • Files kept for up to 30 days
  • MCP server, CLI and REST API
Start free →
Standard
$9.99/ month

For you and every agent you connect.

  • Uploads up to 100 MB
  • 10 GB of storage
  • Links that last as long as you want
  • Private files & expiring links
  • MCP server, CLI and REST API
  • Scoped API keys
  • 50 AI images a month
Start dropping →
Pro
$29.99/ month

For agents that make a lot of images.

  • Everything in Standard
  • Uploads up to 500 MB
  • 100 GB of storage included, then $0.10 per GB a month
  • 200 AI images a month included, then $0.05 each
  • Extras on your monthly invoice
Go Pro →

Upload once.

Then find, use and share it across every AI workflow. Your library is under a minute away.

Get started

Your experience on this site will be improved by allowing cookies.