# API & MCP ClearFeed provides two programmatic ways to manage requests: the **REST API** for direct integration and the **MCP server** for AI agent workflows. *** ## REST API ClearFeed's REST API provides comprehensive access to create, search, update, and analyze requests. Build custom integrations, automate workflows, and connect ClearFeed to your existing tools. ### What You Can Do | Area | Capabilities | | ----------------- | --------------------------------------------------- | | **Requests** | Create, search, update, post messages, close | | **Insights** | Query response times, SLA metrics, team performance | | **Customers** | Look up and update customer information | | **Collections** | List collections, add channels | | **Custom Fields** | List, create, update, delete custom fields | ### Getting Started 1. Generate a **Personal Access Token** from **Settings → Developer Settings** 2. Use the token in the `Authorization: Bearer` header for all API requests 3. See [Developer Settings](/clearfeed-help-center/account-setup/developer-settings.md) for authentication details **Full API Reference:** [ClearFeed API Documentation](https://docs.clearfeed.ai/api/reference) *** ## MCP Server The **MCP (Model Context Protocol) server** exposes ClearFeed as a set of tools that AI assistants like Claude, Cursor, and Codex can call. Ask an assistant to find a request, post a reply, or pull an analytics number, and it talks to ClearFeed for you. ClearFeed offers two ways to connect: * **Remote MCP server** at `https://mcp.clearfeed.app/mcp`. Recommended for most teams; nothing to install or host. * **Local stdio server.** Run the open-source package on your own machine if you prefer to host it yourself or want to embed it in your own server. ### What You Can Do | Area | Example prompts for AI agents | | ------------------- | --------------------------------------------------------- | | **Search requests** | "Find all high-priority requests from the last week" | | **Summarize** | "Summarize the open requests assigned to me" | | **Update requests** | "Mark request CF-1234 as in progress and assign to @john" | | **Post updates** | "Reply to request CF-5678 saying we're investigating" | | **Look up details** | "Show me the customer info for request CF-9012" | | **Analytics** | "What's our average response time this month?" | | **Link tickets** | "Link the Jira ticket PROJ-123 to request CF-4567" | ### Available Tools | Category | Tools | | ----------- | ----------------------------------------------------------------------------------------------------------------- | | Requests | `requests_search`, `requests_list`, `requests_get`, `requests_create`, `requests_update`, `requests_post_message` | | Customers | `customers_list`, `customers_search`, `customers_get`, `customers_create`, `customers_update` | | Collections | `collections_list`, `collections_add_channels` | | Tickets | `tickets_get_form`, `tickets_link` | | Insights | `insights_query` | | Other | Channels, Custom Fields, Teams, Users | *** ### Remote MCP server Add `https://mcp.clearfeed.app/mcp` to your AI assistant once and every user on your team can sign in with their own ClearFeed login. Each user authorizes through OAuth, and ClearFeed handles the API token in the background. #### Setup in Claude Claude supports the remote MCP server through its Custom Connectors. There are two ways to add it. **Option 1: Add it for the whole team (recommended)** A workspace admin can add the connector once and make it available to everyone: 1. Go to . 2. Click **Add custom connector**. 3. Name it **ClearFeed** and set the URL to `https://mcp.clearfeed.app/mcp`. 4. Save.

Adding ClearFeed as a custom connector in Claude

Once the connector is published, every member of the workspace will see it in their connector list and can sign in individually. **Option 2: Add it just for yourself** If you do not have admin access, or you want to try it on your own first: 1. Go to . 2. Click **Add custom connector**. 3. Use the same name and URL: **ClearFeed** and `https://mcp.clearfeed.app/mcp`. #### Signing in After the connector is added, each user has to complete the OAuth flow once before Claude can call any tool: 1. Open Claude and start a conversation. 2. Pick the **ClearFeed** connector. 3. Claude opens a browser window where you sign in to ClearFeed and approve access. Tool calls fail until that flow is finished, so anyone on the team who wants to use ClearFeed in Claude has to do this step on their own login. {% hint style="info" %} **Sign in to the ClearFeed account you want to connect.** The connector is bound to the account you are active on at the moment you authorize. If you belong to multiple accounts, switch to the one you want to connect in ClearFeed before completing the OAuth flow. {% endhint %} #### Connecting more than one account Today a single connector is tied to a single ClearFeed account. If you work across two accounts (for example a parent account and a child account), add the connector twice with different names, like **ClearFeed (Parent)** and **ClearFeed (Child)**, each pointing at `https://mcp.clearfeed.app/mcp`. Sign in to the matching ClearFeed account when you authorize each one. #### Controlling which tools Claude can run Claude lets you decide, per tool, whether it runs automatically, asks for approval first, or is turned off entirely. Read-only tools (search, list, get, analytics) default to **Always allow** and write tools (create, update, link, delete) default to **Needs approval**, but you can change any of them.

Per-tool permissions for the ClearFeed connector in Claude

#### Permissions and tokens Tool calls run against the signed-in user's **Personal Access Token**. If the user already has one, it gets reused; if not, ClearFeed creates one the first time they authorize the connector. There is nothing to set up by hand. Every call still respects that user's ClearFeed permissions. The server exposes the full tool list, but a call only succeeds if the signed-in user can do that action in ClearFeed. For example, a user who cannot delete custom fields will see the `Delete custom field` tool in Claude, but the call will fail. If a user is later removed from the ClearFeed account, the next tool call from their connector fails and the token is revoked. They will need to be added back to the account and re-authorize the connector to use it again. *** ### Local stdio server If you prefer to run the MCP server yourself, the same tools are available as an open-source stdio server. 1. Clone the [ClearFeed MCP repository](https://github.com/clearfeed/clearfeed-mcp). 2. Set `CLEARFEED_API_BASE_URL` and `CLEARFEED_API_TOKEN` environment variables. Generate the token from **Settings → Developer Settings**. 3. Point your MCP client (Claude Code, Cursor, etc.) at the local process. This mode runs with a single API token, so every tool call uses that one user's permissions. **Full Documentation:** [MCP Server Setup](https://docs.clearfeed.ai/api/quick-start/mcp-server) *** ## FAQs 1. **How do I get my API token?**\ **Answer:** Navigate to **Settings → Developer Settings** and click **Generate Token**. See [Developer Settings](/clearfeed-help-center/account-setup/developer-settings.md). 2. **What's the difference between REST API and MCP?**\ **Answer:** REST API is for direct HTTP-based integration. MCP server wraps the REST API as tools for AI agents to use via the Model Context Protocol. 3. **Do I need to create an API token to use the remote MCP server?**\ **Answer:** No, not by hand. The remote server at `mcp.clearfeed.app/mcp` uses OAuth. When you authorize the connector, ClearFeed reuses your existing Personal Access Token or creates one for you automatically. You only need to generate a token yourself if you run the local stdio server. 4. **Can one connector talk to more than one ClearFeed account?**\ **Answer:** Not today. A connector is bound to the ClearFeed account you were signed into when you authorized it. To work with multiple accounts, add the connector again under a different name and sign in to the other account during that OAuth flow. 5. **A tool shows up in Claude but fails when I run it. Why?**\ **Answer:** The MCP server exposes the full tool list to every user, but each call still runs against your ClearFeed permissions. If you do not have access to that feature in ClearFeed, the call will fail. Ask an admin to grant the matching permission. 6. **Are webhooks available for real-time updates?**\ **Answer:** Yes, configure webhooks in **Settings → Developer Settings → Events API Subscription**. --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://docs.clearfeed.ai/clearfeed-help-center/managing-requests/api-and-mcp.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.