# MCP Server

The Archera MCP Server connects your AI assistant to your Archera account using the [Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction), giving it direct access to your cloud commitments, cost data, and optimization recommendations. Archera hosts and manages the remote MCP server with OAuth authentication.

**MCP server URL:** `https://mcp.archera.ai/mcp`

***

## Authentication

The Archera MCP Server uses OAuth Authorization Code with PKCE. MCP clients that support Dynamic Client Registration can register themselves automatically during setup.

Most clients start a browser-based Archera login when you connect to or start the MCP server. Claude Code is different: add the server first, then run `/mcp` inside Claude Code to authenticate.

If your MCP client asks you to manually provide a `client_id`, update the client or contact Archera support.

***

## Set up your client

Select your client below for setup instructions. All clients should connect to `https://mcp.archera.ai/mcp` and authenticate with your existing Archera login.

<details>

<summary>Claude Code</summary>

Run the following command in your terminal:

```bash
claude mcp add --transport http --scope user archera https://mcp.archera.ai/mcp
```

Then open Claude Code and run:

```
/mcp
```

Select the `archera` server and follow the browser login flow. Once you approve access, the server is available in future Claude Code sessions. The `--scope user` flag makes the server available across projects; drop it if you want the server configured only for the current project.

</details>

<details>

<summary>OpenAI Codex</summary>

Run the following command in your terminal:

```bash
codex mcp add archera --url https://mcp.archera.ai/mcp
```

Verify the server is configured:

```bash
codex mcp list
```

Codex shares MCP configuration between the CLI and IDE extension. Start Codex after adding the server and complete the OAuth login flow when prompted.

</details>

<details>

<summary>Claude Desktop</summary>

Open Claude Desktop, then go to **Settings** -> **Connectors**.

1. Click **Add connector** or **+** -> **Add custom connector**.
2. If prompted for a connector type, choose **Custom** -> **Web**.
3. Name the connector `Archera`.
4. Enter the connector URL: `https://mcp.archera.ai/mcp`
5. Click **Add**, then click **Connect** to authenticate with Archera.

Claude Desktop uses the same remote connector flow as Claude on the web. The local Developer config editor is intended for local MCP server processes.

</details>

<details>

<summary>Claude.ai</summary>

1. Go to **Settings** -> **Connectors**.
2. Click **Add connector** or **+** -> **Add custom connector**.
3. If prompted for a connector type, choose **Custom** -> **Web**.
4. Name the connector `Archera`.
5. Enter the connector URL: `https://mcp.archera.ai/mcp`
6. Click **Add**, then click **Connect** to authenticate with Archera.

For Team or Enterprise plans, an owner may need to add the connector from **Organization settings** -> **Connectors** before members can connect it.

</details>

<details>

<summary>Cursor</summary>

Go to **Cursor** -> **Settings** -> **Cursor Settings** -> **MCP** and add the server. You can also edit your `mcp.json` directly:

```json
{
  "mcpServers": {
    "archera": {
      "type": "http",
      "url": "https://mcp.archera.ai/mcp"
    }
  }
}
```

Use `~/.cursor/mcp.json` for a global server or `.cursor/mcp.json` for a project-specific server. Start or refresh the server in Cursor, then follow the OAuth prompt to connect your Archera account.

</details>

<details>

<summary>Gemini CLI</summary>

Run the following command in your terminal:

```bash
gemini mcp add archera https://mcp.archera.ai/mcp --transport http --scope user
```

Start Gemini CLI. If authentication does not start automatically, run:

```
/mcp auth archera
```

The `--scope user` flag makes the server available across projects; use `--scope project` if you want the server configured only for the current project.

</details>

<details>

<summary>Cline</summary>

Open the Cline panel in VS Code, then open **MCP Servers** from the panel menu.

1. Select the **Remote Servers** tab.
2. Enter server name `archera`.
3. Enter server URL `https://mcp.archera.ai/mcp`.
4. Select **Streamable HTTP** as the transport type.
5. Click **Add Server**.
6. If Cline shows **Authentication required**, click **Authenticate** and complete the browser login flow.

For advanced setup, add the server to Cline's MCP config:

```json
{
  "mcpServers": {
    "archera": {
      "url": "https://mcp.archera.ai/mcp",
      "type": "streamableHttp",
      "disabled": false
    }
  }
}
```

</details>

<details>

<summary>Zed</summary>

Open the Agent Panel's settings view and click **Add Custom Server**, or add the server directly to your Zed settings:

```json
{
  "context_servers": {
    "archera": {
      "url": "https://mcp.archera.ai/mcp"
    }
  }
}
```

Zed prompts for OAuth authentication when the remote MCP server has no configured `Authorization` header.

</details>

<details>

<summary>VS Code (GitHub Copilot)</summary>

Open the Command Palette with `CMD+Shift+P`, select **MCP: Add Server**, choose **HTTP**, and enter:

```
https://mcp.archera.ai/mcp
```

Name the server `archera`, then choose whether to save it globally in your user profile or in the current workspace.

You can also open your user MCP configuration and add:

```json
{
  "servers": {
    "archera": {
      "type": "http",
      "url": "https://mcp.archera.ai/mcp"
    }
  }
}
```

Start the server, confirm that you trust it if prompted, and complete the OAuth login flow.

</details>

<details>

<summary>Windsurf</summary>

Open the Cascade panel, select the **MCPs** icon, or go to **Windsurf Settings** -> **Cascade** -> **MCP Servers**. Add a custom MCP server with this configuration:

```json
{
  "mcpServers": {
    "archera": {
      "serverUrl": "https://mcp.archera.ai/mcp"
    }
  }
}
```

If you edit the raw config directly, it is stored in `~/.codeium/windsurf/mcp_config.json`. Start or refresh the server, then follow the OAuth prompt to connect your Archera account.

</details>

<details>

<summary>Warp</summary>

Go to **Settings** -> **Agents** -> **MCP servers** -> **+ Add** and select **Streamable HTTP or SSE Server (URL)**. Enter the following:

```json
{
  "mcpServers": {
    "archera": {
      "url": "https://mcp.archera.ai/mcp"
    }
  }
}
```

Start the server. If you have not authenticated before, Warp opens a browser window to authenticate with Archera.

</details>

<details>

<summary>Amp</summary>

**VS Code Extension:** Add the server via the Amp VS Code extension settings, or update your `settings.json`:

```json
"amp.mcpServers": {
  "archera": {
    "url": "https://mcp.archera.ai/mcp"
  }
}
```

**Amp CLI:**

```bash
amp mcp add archera https://mcp.archera.ai/mcp
```

Amp starts the OAuth flow in your browser when the remote server starts. If the token becomes stale, run `amp mcp oauth logout archera` and restart Amp to authenticate again.

</details>

<details>

<summary>Other clients</summary>

The Archera MCP server works with clients that support Streamable HTTP, OAuth Authorization Code with PKCE, and Dynamic Client Registration. Use `https://mcp.archera.ai/mcp` as the server URL in your client's MCP configuration.

Discovery endpoints:

```
Protected resource metadata:
https://mcp.archera.ai/.well-known/oauth-protected-resource/mcp

Authorization server metadata:
https://api.archera.ai/.well-known/oauth-authorization-server

Dynamic client registration:
https://api.archera.ai/oauth/register
```

</details>

***

## Example prompts

Once connected, try these prompts to get started:

**Visibility**

* What commitment plans do I have active right now?
* Show me my Reserved Instance and Savings Plan coverage across all AWS accounts
* Which of my commitments are expiring in the next 60 days?
* What's my current on-demand spend versus committed spend?

**Analysis**

* Where am I most over- or under-committed?
* Show me EC2 utilization and coverage for the past 30 days
* Which accounts have the lowest commitment coverage?
* Are there any unused reservations I should be aware of?

**Optimization**

* What savings opportunities does Archera see for my environment?
* Generate a commitment plan recommendation for my production AWS account
* How much could I save by converting my top on-demand spend to Savings Plans?
* Compare my commitment efficiency this quarter versus last quarter

***

## Available tools

The Archera MCP server gives your AI assistant access to:

**Commitment management** View and manage your full portfolio of Reserved Instances, Savings Plans, and Archera Insured Commitments across all linked cloud accounts and services. Check status, term lengths, payment options, and renewal dates.

**Coverage & utilization** Analyze how well your existing commitments are being used. Identify underutilized reservations, coverage gaps by service or region, and accounts where on-demand spend is higher than it should be.

**Recommendations & planning** Generate commitment plans and savings recommendations based on your actual usage patterns. Model different scenarios, including term length, payment type, and coverage targets, to find the right balance of savings and flexibility for your organization.

**Spend & forecasting** Query historical cost data broken down by account, service, region, or commitment type. Get projected spend estimates based on current commitments and usage trends.

**Account management** View linked cloud accounts, their coverage summaries, and commitment activity across your Archera organization.

***

## Troubleshooting

**Auth not completing in Claude Code** Run `/mcp` inside Claude Code, select the `archera` server, and choose the authentication option. If the browser does not open automatically, copy the URL from the terminal and paste it manually. If your client uses a local callback server, make sure the callback port is not blocked by a firewall or VPN.

**Client asks for a client ID** Make sure your MCP client supports OAuth Dynamic Client Registration. If the client still requires a manually issued client ID, update the client or contact <support@archera.ai>.

**Reconnecting or switching accounts** In Claude Code, type `/mcp`, select the `archera` server, and choose **Clear authentication**. Then re-run the setup command to trigger a fresh login.

**Missing tools or data** Ensure your Archera account has the necessary permissions and that authentication completed successfully. Try disconnecting and reconnecting the MCP server.

**Connection issues** Verify the server URL is correct: `https://mcp.archera.ai/mcp`. Check that your network can reach `mcp.archera.ai` and `api.archera.ai`.

For further help, contact <support@archera.ai>.


---

# 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.archera.ai/api-reference/mcp-server/mcp.md?ask=<question>
```

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.