Connect Kinn to an AI tool
Kinn supports the Model Context Protocol (MCP), a standard that lets compatible AI assistants use Kinn as a data source and as an analyst. This is useful if your team already works in an AI tool that supports remote MCP servers.
This is an optional connection. You do not need it to use Kinn chat, integrations, dashboards, reports, or exports.
For what the connection can actually do and how to get good results from it, see The Kinn MCP. This page covers connecting it and the permissions involved.
What the connected tool can do
During authorization you choose one Kinn account. The AI tool then gets the same role-dependent capabilities you have in Kinn chat — no more, and no less.
This connection is not read-only. Depending on your role, a connected AI tool can make changes in Kinn and in your product's connected trusted connectors. Kinn only makes changes when they're explicitly requested, but authorize only tools you trust.
Your existing Kinn role decides what's possible:
| Your role | What a connected AI tool can do |
|---|---|
| Viewer | Read and analyze — query feedback, run deep research, generate charts, create reports and exports. |
| Editor | Everything above, plus changes: saving notes, connecting and managing sources, updating product info, importing data, and writing to connected trusted connectors that accept writes (Jira, Linear). |
| Admin | Everything above, across the account. |
Two things are worth being precise about:
- Your role is checked on every call, not just at authorization. If your role changes in Kinn, the connection's capabilities change with it.
- The token covers every product in the account you selected, which may include private community or support content.
How to connect
The exact labels differ by AI tool, but the flow is generally:
- Open the tool's settings for Connectors, Integrations, or MCP.
- Add Kinn as a remote MCP server using
https://mcp.kinn.gg/mcp. - If the tool asks for a transport type, choose Streamable HTTP.
- Sign in to Kinn when your browser opens.
- Review the requesting tool's name and the permissions shown.
- Choose the Kinn account to share and select Allow access.
The AI tool can then list the products in that account and start asking questions. If the account has a single product, it can skip product selection entirely.
Choosing a product
With one product in the account, nothing to do — Kinn uses it automatically.
With several, the AI tool calls list_products (or reads the kinn://products
resource) and passes the product ID it wants. In practice you can just name the
product in your request and let the assistant resolve it.
Usage is billed to whichever product is selected, the same as a chat conversation.
Continuing a conversation
Every call creates a durable Kinn thread and returns its ID. Well-behaved AI tools pass that ID back automatically so follow-up questions continue the same investigation rather than starting over.
Note that these threads live only in MCP — they don't appear in your Kinn web chat history, and can't be continued from the web UI.
Security notes
- Only authorize tools you trust. The connection can read feedback for every product in the selected account, and — depending on your role — make changes.
- Product descriptions are treated as untrusted data. Kinn never follows instructions embedded in a product's description or in the feedback it reads. Neither should your assistant.
- Revoke from your AI tool to end access. Reconnecting requires signing in again.
- Use a Viewer account for read-only setups. If you want an assistant that can analyze but never change anything, connect it with a Viewer-role user.
If the connection does not work
- Confirm the server address is exactly
https://mcp.kinn.gg/mcp. - Confirm the transport is Streamable HTTP.
- Make sure you belong to at least one Kinn account.
- Remove the connection in your AI tool and authorize it again.
- Ask your AI tool's administrator whether remote MCP servers are allowed for your workspace.
- If file attachments aren't working — they're not supported over MCP. Use Kinn chat for those.
Next
- The Kinn MCP — what's exposed, what makes it different, and how to get good answers through it.