About Extend
Extend Agent Plugin packages document-processing tools and integration skills for agents. It connects to Extend’s hosted MCP server for parsing, extraction, classification, splitting, PDF editing, and workflows across supported document types.
Key features
- 0180+ hosted MCP tools
- 02Document parsing and field extraction
- 03Classification and bundle splitting
- 04PDF editing and form filling
- 05API and CLI agent skills
Use cases
- 01Extract structured fields from documents
- 02Classify documents and split bundles
- 03Fill or edit PDF files
- 04Build Extend API integrations or CLI workflows
MCP support
The plugin connects to Extend’s hosted MCP server at `https://mcp.extend.ai/mcp` using Streamable HTTP and OAuth. The server exposes tools for document processing, workflows, files, webhooks, and evaluations.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://mcp.extend.ai/mcpclaude mcp add --transport http extend https://mcp.extend.ai/mcp{"mcpServers":{"extend":{"type":"http","url":"https://mcp.extend.ai/mcp"}}}Open in Cursor{"servers":{"extend":{"type":"http","url":"https://mcp.extend.ai/mcp"}}}Open in VS Code{"mcpServers":{"extend":{"type":"http","url":"https://mcp.extend.ai/mcp"}}}{"mcpServers":{"extend":{"serverUrl":"https://mcp.extend.ai/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEExtend Agent Plugin
Document processing for agents, packaged as a plugin. Extend turns unstructured documents (PDFs, images, Office files, spreadsheets) into structured data: parsing, field extraction, classification, splitting, PDF form filling, and multi-step document workflows.
This plugin bundles:
- The hosted Extend MCP server (
https://mcp.extend.ai/mcp): 80+ tools covering extraction, parsing, classification, splitting, PDF editing, workflows, files, webhooks, and evaluations. Sign-in is OAuth: your client opens a browser window where you choose the workspace and environment the connection may use. The server teaches its own cross-tool usage patterns through MCP server instructions and per-response guidance, so no companion skill is needed for it. extend-apiskill: a compact brief for agents writing integration code against the Extend API and SDKs: endpoints, schema rules, sync-vs-async patterns, webhooks, and error handling.extend-cliskill: how to drive theextendCLI for shell, script, and CI work against local files.
The skills also install standalone, without the plugin:
npx skills add extend-hq/extend-agent-plugin
Install
- Cursor: install "Extend" from the Cursor Marketplace,
or load this repository from
~/.cursor/plugins/local/for development. - Claude Code:
claude plugin install extendonce listed in the plugin directory, or add this repository as a plugin marketplace. - Any MCP client: you don't need the plugin to use Extend; connect
directly to
https://mcp.extend.ai/mcp(Streamable HTTP + OAuth). See the connection guide for per-client instructions.
Repository layout
Several manifest formats coexist so one repository serves every ecosystem:
| Path | Purpose |
|---|---|
plugin.json, mcp.json | Agent Plugins standard (any conformant client) |
.cursor-plugin/plugin.json | Cursor plugin manifest (marketplace metadata and logo; shares mcp.json) |
.claude-plugin/plugin.json, .mcp.json | Claude plugin manifest |
gemini-extension.json | Gemini CLI extension manifest |
server.json | Official MCP Registry entry (ai.extend/extend) |
skills/ | Agent Skills shared by every format |
assets/ | Logo referenced by the manifests |
Privacy and terms
The MCP server calls the Extend API on your behalf under the workspace and environment you approve at sign-in. See Extend's privacy policy, terms, and security overview. Support: [email protected].
License
MIT; see LICENSE. The license covers this plugin packaging; the Extend platform and hosted MCP server are Extend's commercial services.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.