# Weav [Health: Active]

**Category:** 🎧 Support & Service Management  
**Repository:** https://github.com/dubdubdubco/weav-mcp-server  
**GitHub Stars:** 0  
**Views:** 1  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/weav

## Description
Connect AI tools to Weav with MCP. Search conversations, reply to customers, and manage knowledge.

## Tools
Capabilities this server exposes over MCP:

- **status** — `open` or `closed`
- **priority** — `low`, `normal`, `high`, or `urgent`
- **type** — `text`, `qa`, or `url`

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "weav": {
    "url": "https://mcp.weav.com/mcp"
  }
}
```

## Documentation

## What Weav MCP server does

The Weav MCP server gives MCP-compatible AI applications access to a Weav workspace. Its 32 tools cover four areas: conversations, customers, agent training data, and the public Help Center knowledge base.

Conversation tools can search inbox content, retrieve recent messages, send replies as a workspace teammate, and change status, priority, or assignment. Customer tools search by name or email and return profiles, company information, and recent conversations.

Training-data tools work with the material used by Weav AI agents. They can find, retrieve, add, delete, and resync items. New sources may be text, question-and-answer content, or website URLs. Help Center tools cover categories, articles, translations, revisions, improvement proposals, publishing state, and soft deletion.

## How it works

Weav hosts the remote MCP service at `https://mcp.weav.com/mcp` using Streamable HTTP. An AI client connects to that endpoint, then a Weav workspace administrator or owner completes OAuth consent in a browser. Once authorized, the client can call tools against the connected workspace until the authorization is revoked.

The workspace server documented here is different from Weav's unauthenticated public server at `https://weav.com/mcp`. The public server answers pricing and product questions but cannot access inboxes, customers, training data, or Help Center data.

## Setup and configuration

Configure an MCP client with the Streamable HTTP URL:

```text
https://mcp.weav.com/mcp
```

Authentication is handled through OAuth, and consent must be completed by a workspace admin or owner. After connection, the client can retain access until the authorization is revoked.

## Tools and capabilities

The Weav MCP server exposes these functional groups:

- Conversation search and retrieval, customer replies, and conversation updates.
- Customer search and profile retrieval.
- Agent training-data search, retrieval, creation, deletion, and URL resynchronization.
- Help Center overview, category and article listing, article search, and full article retrieval.
- Article revision listing, retrieval, and restoration.
- Read-only improvement-proposal listing and retrieval.
- Help Center article creation, updates, publishing, unpublishing, soft deletion, and translation deletion.

Many list and search operations support pagination or limits from 1 to 50. Conversation searches can filter by status and channel, while Help Center searches can filter by locale, publication status, category, or uncategorized state. Article changes may use an expected version, and editing existing translation content creates a revision.

## Limitations and notes

Access is limited to the authorized Weav workspace and depends on the permissions granted through OAuth. Sending a reply changes a real conversation, and training-data additions are automatically available to every agent in the workspace. Deleting URL training data also removes its discovered URLs and chunks.

Agent training data and Help Center content are separate systems. The `*_training_data` tools manage sources used by Weav AI agents; the `kb_*` tools manage public Help Center articles and categories. Conversation updates do not delete conversations, and Help Center article deletion is described as a soft delete.

_Full upstream README: https://allmcps.com/mcp/weav/readme_