Connect Claude, Claude Code, ChatGPT, Cursor, VS Code or Codex to your AmICited workspace over MCP. One URL, OAuth sign-in, and an API key as a fallback.
AmICited runs an always-on MCP (Model Context Protocol) server. Once your AI client is connected, it can read your visibility data, run reports and, with write access, add prompts, generate articles or log annotations, all from a normal conversation. Every dashboard feature also ships as an MCP tool, so there is nothing you can do in the app that an agent cannot.
A plan that includes MCP. MCP is available on the 7-day trial and on Pro, Premium and Enterprise. Starter workspaces can see the server URL but connections are refused until the plan is upgraded. See plans and limits.
Editor permissions or higher in the workspace. The MCP page tells members without them to ask a workspace owner or admin.
A positive credit balance. Reads over MCP are free, but the endpoint checks that the workspace has credits before it answers. See credits.
The workspace is not in the URL. It comes from your credential: with OAuth you pick the workspace on the consent screen, and an API key already belongs to one workspace. You can also copy the URL from the MCP page in the app, which shows it with a Copy URL button and the same per-client instructions as below. The transport is streamable HTTP.
Paste https://api.flowhunt.io/mcp/amicited and click Add.
Click Connect on the new connector and approve access in the browser window that opens. Choose your workspace on the consent screen.
In a chat, open the tools menu and make sure the AmICited connector is enabled.
API key alternative (Claude Desktop config file). Add this under mcpServers in ~/Library/Application Support/Claude/claude_desktop_config.json on macOS, then restart Claude Desktop:
Custom MCP connectors need Developer mode, which is available on paid ChatGPT plans (Plus, Pro, Team, Enterprise). In Business, Enterprise and Edu workspaces an admin may need to enable it for you.
In ChatGPT, open Settings, then Apps & Connectors.
Add a custom connector and paste https://api.flowhunt.io/mcp/amicited.
Approve access in the browser window that opens.
API key alternative. At step 4, choose API key authentication instead and paste your workspace API key. Save the connector, then enable it in a new chat.
Cursor
On the app’s MCP page, select Cursor and click Add to Cursor. Cursor installs the server and runs the sign-in for you.
If the button does not open Cursor, add the server by hand in Cursor’s settings under MCP, or edit ~/.cursor/mcp.json (or a per-project .cursor/mcp.json):
API key alternative. Add "headers": { "Authorization": "Bearer <your-workspace-api-key>" } to the entry. You can also run MCP: Add Server…, choose HTTP (HTTP or Server-Sent Events), paste the URL and add the Authorization header when prompted.
Codex
Codex connects as a standard remote MCP client. Add the server, then sign in:
Paste the URL plus an Authorization: Bearer header
Workspace
Chosen on the consent screen
Fixed by the key
Permissions
amicited:read to connect and read; amicited:write also needed for writes
Full workspace access
Revoking
Connected clients → Revoke on the MCP page
Delete or expire the key
OAuth is the better default: there is no secret to paste into a config file, each client shows up by name on the MCP page under Connected clients, and you can revoke one without touching the others. A read-only OAuth token can list and read everything, but any tool that creates, edits or deletes will refuse with a message naming the amicited:write scope. Retrying will not help; reconnect with write access.
Treat an API key like a password
Anyone holding a workspace API key can run every tool on the server against your workspace data, including writes. Prefer OAuth, keep keys out of shared repositories, and use an environment variable where your client supports one.
Ask your agent something small: “List my AmICited domains.” It should call list_domains and return your domains by name. Then try “Show my AI visibility for the last 30 days”, which uses prompt_analytics.
If the connection is refused, the message usually says why:
Plan. “requires a Pro plan or above”: the workspace owner’s plan does not include MCP.
Credits. The workspace balance is zero or negative. Top up and retry.
Rate limit. Tool calls are capped per workspace per hour: 120 on the trial, 600 on Pro, 2,000 on Premium, unlimited on Enterprise. The message says when the window resets. Listing tools and reading skills do not count.
# Connect the MCP server
Connect Claude, Claude Code, ChatGPT, Cursor, VS Code or Codex to your AmICited workspace over MCP. One URL, OAuth sign-in, and an API key as a fallback.
AmICited runs an always-on MCP (Model Context Protocol) server. Once your AI client is connected, it can read your visibility data, run reports and, with write access, add prompts, generate articles or log annotations, all from a normal conversation. Every dashboard feature also ships as an MCP tool, so there is nothing you can do in the app that an agent cannot.
## Before you start
- **A plan that includes MCP.** MCP is available on the 7-day trial and on Pro, Premium and Enterprise. Starter workspaces can see the server URL but connections are refused until the plan is upgraded. See [plans and limits](/docs/account/plans-and-limits/).
- **Editor permissions or higher** in the workspace. The **MCP** page tells members without them to ask a workspace owner or admin.
- **A positive credit balance.** Reads over MCP are free, but the endpoint checks that the workspace has credits before it answers. See [credits](/docs/account/credits/).
## The server URL
There is one URL for everyone:
```text
https://api.flowhunt.io/mcp/amicited
```
The workspace is not in the URL. It comes from your credential: with OAuth you pick the workspace on the consent screen, and an API key already belongs to one workspace. You can also copy the URL from the **MCP** page in the app, which shows it with a **Copy URL** button and the same per-client instructions as below. The transport is streamable HTTP.
## Connect your client
Pick your client. Each tab starts with the recommended OAuth sign-in and ends with the API key alternative.
Connectors you add on claude.ai are also available in the Claude Desktop app.
1. Open [Claude's connector settings](https://claude.ai/customize/connectors).
2. Click **+**, then **Add custom connector**.
3. Paste `https://api.flowhunt.io/mcp/amicited` and click **Add**.
4. Click **Connect** on the new connector and approve access in the browser window that opens. Choose your workspace on the consent screen.
In a chat, open the tools menu and make sure the AmICited connector is enabled.
**API key alternative (Claude Desktop config file).** Add this under `mcpServers` in `~/Library/Application Support/Claude/claude_desktop_config.json` on macOS, then restart Claude Desktop:
```json {title="claude_desktop_config.json"}
{
"mcpServers": {
"amicited": {
"type": "http",
"url": "https://api.flowhunt.io/mcp/amicited",
"headers": {
"Authorization": "Bearer <your-workspace-api-key>"
}
}
}
}
```
Register the server once from your terminal:
```bash
claude mcp add --transport http amicited https://api.flowhunt.io/mcp/amicited
```
Then run `/mcp` inside Claude Code, choose **Authenticate**, and approve access in the browser window that opens.
**API key alternative.** Add the server under the top-level `mcpServers` key in `~/.claude.json` and restart Claude Code:
```json {title="~/.claude.json"}
{
"mcpServers": {
"amicited": {
"type": "http",
"url": "https://api.flowhunt.io/mcp/amicited",
"headers": {
"Authorization": "Bearer <your-workspace-api-key>"
}
}
}
}
```
Custom MCP connectors need **Developer mode**, which is available on paid ChatGPT plans (Plus, Pro, Team, Enterprise). In Business, Enterprise and Edu workspaces an admin may need to enable it for you.
1. Turn on Developer mode as described in [OpenAI's connector documentation](https://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt).
2. In ChatGPT, open **Settings**, then **Apps & Connectors**.
3. Add a custom connector and paste `https://api.flowhunt.io/mcp/amicited`.
4. Approve access in the browser window that opens.
**API key alternative.** At step 4, choose API key authentication instead and paste your workspace API key. Save the connector, then enable it in a new chat.
On the app's **MCP** page, select **Cursor** and click **Add to Cursor**. Cursor installs the server and runs the sign-in for you.
If the button does not open Cursor, add the server by hand in Cursor's settings under MCP, or edit `~/.cursor/mcp.json` (or a per-project `.cursor/mcp.json`):
```json {title="~/.cursor/mcp.json"}
{
"mcpServers": {
"amicited": {
"url": "https://api.flowhunt.io/mcp/amicited"
}
}
}
```
Cursor opens a browser to sign in the first time it connects. Set the chat to **Agent** mode to use the tools.
**API key alternative.** Add a header to the same entry:
```json
"headers": { "Authorization": "Bearer <your-workspace-api-key>" }
```
VS Code uses the server through **GitHub Copilot in Agent mode**, so Copilot must be enabled.
On the app's **MCP** page, select **VS Code** and click **Add to VS Code**. VS Code installs the server and runs the sign-in for you.
To add it by hand, open the Command Palette, run **MCP: Open User Configuration**, and add the server under `servers`:
```json {title="mcp.json"}
{
"servers": {
"amicited": {
"type": "http",
"url": "https://api.flowhunt.io/mcp/amicited"
}
}
}
```
**API key alternative.** Add `"headers": { "Authorization": "Bearer <your-workspace-api-key>" }` to the entry. You can also run **MCP: Add Server...**, choose **HTTP (HTTP or Server-Sent Events)**, paste the URL and add the `Authorization` header when prompted.
Codex connects as a standard remote MCP client. Add the server, then sign in:
```bash
codex mcp add amicited --url https://api.flowhunt.io/mcp/amicited
codex mcp login amicited
```
Or edit `~/.codex/config.toml` directly:
```toml {title="~/.codex/config.toml"}
[mcp_servers.amicited]
url = "https://api.flowhunt.io/mcp/amicited"
oauth_resource = "https://api.flowhunt.io/mcp/amicited"
```
**API key alternative.** Put the key in an environment variable and point Codex at it:
```toml {title="~/.codex/config.toml"}
[mcp_servers.amicited]
url = "https://api.flowhunt.io/mcp/amicited"
bearer_token_env_var = "AMICITED_API_KEY"
```
Codex runs with strict approval settings will still reach every read-only tool through `read_tool`. See [toolsets and tools](/docs/mcp/toolsets/).
## OAuth or API key
| | OAuth 2.1 (recommended) | Workspace API key (fallback) |
|---|---|---|
| **Setup** | Paste the URL, approve in the browser | Paste the URL plus an `Authorization: Bearer` header |
| **Workspace** | Chosen on the consent screen | Fixed by the key |
| **Permissions** | `amicited:read` to connect and read; `amicited:write` also needed for writes | Full workspace access |
| **Revoking** | **Connected clients → Revoke** on the MCP page | Delete or expire the key |
OAuth is the better default: there is no secret to paste into a config file, each client shows up by name on the **MCP** page under **Connected clients**, and you can revoke one without touching the others. A read-only OAuth token can list and read everything, but any tool that creates, edits or deletes will refuse with a message naming the `amicited:write` scope. Retrying will not help; reconnect with write access.
Anyone holding a workspace API key can run every tool on the server against your workspace data, including writes. Prefer OAuth, keep keys out of shared repositories, and use an environment variable where your client supports one.
## Check that it works
Ask your agent something small: *"List my AmICited domains."* It should call `list_domains` and return your domains by name. Then try *"Show my AI visibility for the last 30 days"*, which uses `prompt_analytics`.
If the connection is refused, the message usually says why:
- **Plan.** "requires a Pro plan or above": the workspace owner's plan does not include MCP.
- **Credits.** The workspace balance is zero or negative. Top up and retry.
- **Rate limit.** Tool calls are capped per workspace per hour: 120 on the trial, 600 on Pro, 2,000 on Premium, unlimited on Enterprise. The message says when the window resets. Listing tools and reading skills do not count.
## Related
How an agent finds the right tool among 300+ without loading them all.Copy-paste prompts for briefings, drops, gaps and articles.The built-in playbooks agents read before multi-step questions.Which plans include MCP and what each allows.
Source: https://www.amicited.com/docs/mcp/connect/
Esc
↑↓ navigate↵ open
Cookie Consent We use cookies to enhance your browsing experience and analyze our traffic. Privacy Policy.
Cookie Settings
Necessary Cookies
These cookies are required for the website to function and cannot be disabled.
Analytics Cookies
These cookies help us understand how visitors interact with our website.
Marketing Cookies
These cookies are used to measure the effectiveness of our advertising campaigns and to support relevant advertising on third-party platforms.
Functional Cookies
These cookies are used to remember your preferences and improve your experience.
