Your agent,connected.

Claude.ai and Claude Desktop

Web formAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

Custom connectors work on Free (one connector), Pro, Max, Team and Enterprise. On Team and Enterprise an Owner adds it once for everyone.

1. Open Customize > Connectors and click Add custom connector. (Team or Enterprise Owner: Organization settings > Connectors > Add > Custom > Web.)
2. Name it ActionLake and paste the MCP server URL: https://connect.actionlake.com/mcp.
3. Leave the OAuth client fields empty. If the dialog asks how Claude identifies itself, choose Register automatically.
4. Click Add, then Connect. Sign in on the ActionLake page, or create a free account there with one click.
5. Prefer a key? If your dialog shows Request headers (beta), choose No sign-in and add the header x-api-key with your key as the value.

https://connect.actionlake.com/mcp

x-api-key: al_YOUR_KEY

Fill in a key for Claude.ai and Claude Desktop →

ChatGPT

Web formsign-inNot yet tested by us; from the vendor's docs, 2026-10-22

ChatGPT connects with sign-in only; it does not send API keys. Full read and write needs ChatGPT Business, Enterprise or Edu on the web. Pro can connect for read and fetch only. In a Business workspace only admins and owners can add one; on Enterprise and Edu an admin grants Create plugins with MCPs.

1. Open chatgpt.com/plugins, click +, and choose Add custom MCP server.
2. Name it ActionLake. Connection: the URL, including /mcp: https://connect.actionlake.com/mcp.
3. Authentication: OAuth.
4. Tick I understand and want to continue, then click Create as a plugin. Review the tools and install it.
5. Sign in on the ActionLake page, or create a free account there. Then mention @ActionLake in a chat.
6. ChatGPT asks you to confirm each action that can change something. Keep that on; it is the safe default.

https://connect.actionlake.com/mcp

Fill in a key for ChatGPT →

Grok

Web formAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

The xAI API has no approval step: the model calls tools on its own. Give it a key with narrow grants.

1. Grok app: open grok.com/connectors, click New Connector, then Custom.
2. Paste https://connect.actionlake.com/mcp and sign in to ActionLake, or create a free account there. On mobile it is Settings > Connectors.
3. Grok CLI: run the command below.
4. xAI API: add an mcp tool. The authorization value can be the bare key.

grok mcp add --transport http actionlake https://connect.actionlake.com/mcp --header "Authorization: Bearer al_YOUR_KEY"

import os
from openai import OpenAI

client = OpenAI(base_url="https://api.x.ai/v1", api_key=os.environ["XAI_API_KEY"])
response = client.responses.create(
    model="grok-4",
    tools=[{
        "type": "mcp",
        "server_label": "actionlake",
        "server_url": "https://connect.actionlake.com/mcp",
        "authorization": "al_YOUR_KEY",
    }],
    input="Use actionlake: call help, then tell me what you can do.",
)
print(response.output_text)

Fill in a key for Grok →

Claude Code

One commandAPI keysign-inVerified in ActionLake's code

1. Run the command once. Add --scope user before the name to use ActionLake in every project.
2. Signing in instead of using a key? Leave the header off, start claude and run /mcp; the ActionLake consent page opens, and it can create a free account there.
3. Ask your agent to call help. The eight tools are listed under actionlake.

claude mcp add --transport http actionlake https://connect.actionlake.com/mcp --header "Authorization: Bearer al_YOUR_KEY"

claude mcp add --transport http actionlake https://connect.actionlake.com/mcp

Fill in a key for Claude Code →

Codex

Config fileAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. Add the table to ~/.codex/config.toml (or .codex/config.toml in a trusted project).
2. Put the key in the environment variable the table names. Codex sends it as Authorization: Bearer.
3. Signing in instead? Run the two commands under sign in; the browser opens the ActionLake consent page.

[mcp_servers.actionlake]
url = "https://connect.actionlake.com/mcp"
bearer_token_env_var = "ACTIONLAKE_KEY"

export ACTIONLAKE_KEY="al_YOUR_KEY"

codex mcp add actionlake --url https://connect.actionlake.com/mcp
codex mcp login actionlake

Fill in a key for Codex →

VS Code and GitHub Copilot

One-click installAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

On Copilot Business or Enterprise, an admin must enable the MCP servers in Copilot policy.

1. Click Install in VS Code. It adds the server with sign-in; VS Code opens the ActionLake consent page the first time a tool needs it.
2. Using a key instead? Put the JSON in .vscode/mcp.json (or your user mcp.json). VS Code asks for the key once and stores it securely, so the file is safe to commit.
3. Open Copilot Chat in Agent mode and ask it to call help.

{
  "inputs": [
    {
      "type": "promptString",
      "id": "actionlake-key",
      "description": "ActionLake API key",
      "password": true
    }
  ],
  "servers": {
    "actionlake": {
      "type": "http",
      "url": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer ${input:actionlake-key}"
      }
    }
  }
}

al_YOUR_KEY

code --add-mcp '{"name":"actionlake","type":"http","url":"https://connect.actionlake.com/mcp"}'

Fill in a key for VS Code and GitHub Copilot →

GitHub Copilot CLI

One commandAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. Run the command. It writes the server to ~/.copilot/mcp-config.json.
2. Signing in instead? Leave the header off; Copilot CLI registers itself and opens the ActionLake consent page.
3. In a session, /mcp lists the server and its tools.

copilot mcp add --transport http --header "Authorization: Bearer al_YOUR_KEY" actionlake https://connect.actionlake.com/mcp

copilot mcp add --transport http actionlake https://connect.actionlake.com/mcp

Fill in a key for GitHub Copilot CLI →

GitHub Copilot cloud agent

Config fileAPI keyNot yet tested by us; from the vendor's docs, 2026-10-22

The cloud agent does not support sign-in (OAuth) for remote servers, and it runs tools without asking. Give it a key of its own with read-only or narrow grants.

1. In the repository, open Settings > Agents secrets (organization or repository) and add a secret named COPILOT_MCP_ACTIONLAKE_KEY with your key as the value.
2. Open Settings > Copilot > MCP servers (under Code, planning, and automation).
3. Paste the JSON into MCP configuration and click Save MCP configuration.
4. tools is required. ["*"] enables all eight; list names to allow fewer.

{
  "mcpServers": {
    "actionlake": {
      "type": "http",
      "url": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer $COPILOT_MCP_ACTIONLAKE_KEY"
      },
      "tools": [
        "*"
      ]
    }
  }
}

al_YOUR_KEY

Fill in a key for GitHub Copilot cloud agent →

Cursor

One-click installAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. Click Add to Cursor. Cursor installs the server with sign-in and opens the ActionLake consent page when you connect.
2. Using a key instead? Put the JSON in ~/.cursor/mcp.json (every project) or .cursor/mcp.json (this one). Keep a file with a key in it out of git.
3. In Cursor Settings > MCP, check that actionlake is on and lists eight tools.

{
  "mcpServers": {
    "actionlake": {
      "url": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer al_YOUR_KEY"
      }
    }
  }
}

Fill in a key for Cursor →

Windsurf

Config fileAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. Open the MCP config: ~/.codeium/windsurf/mcp_config.json (Devin Desktop: ~/.config/devin/mcp_config.json).
2. Add the server. Windsurf reads serverUrl for a remote server.
3. Signing in instead? Leave out headers, refresh the MCP list, and Windsurf opens the ActionLake consent page.

{
  "mcpServers": {
    "actionlake": {
      "serverUrl": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer al_YOUR_KEY"
      }
    }
  }
}

Fill in a key for Windsurf →

Cline

Config fileAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. Open MCP Servers > Configure and edit cline_mcp_settings.json (CLI: ~/.cline/data/settings/cline_mcp_settings.json).
2. "type": "streamableHttp" is required. Without it Cline falls back to the old SSE transport and the connection fails.
3. Or use the CLI command below.

{
  "mcpServers": {
    "actionlake": {
      "type": "streamableHttp",
      "url": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer al_YOUR_KEY"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

cline mcp add actionlake https://connect.actionlake.com/mcp --transport http --header "Authorization: Bearer al_YOUR_KEY" --yes

Fill in a key for Cline →

Zed

Config fileAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. Run zed: open settings file and add the server under context_servers.
2. Signing in instead? Leave out headers; Zed runs the MCP sign-in itself and keeps the token in your keychain.
3. Settings > AI > MCP Servers shows a green dot when it is connected.

{
  "context_servers": {
    "actionlake": {
      "url": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer al_YOUR_KEY"
      }
    }
  }
}

Fill in a key for Zed →

Kiro

One-click installAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. Click Add to Kiro. Kiro installs the server with sign-in and opens the ActionLake consent page.
2. Using a key instead? Put the JSON in ~/.kiro/settings/mcp.json (or .kiro/settings/mcp.json in a workspace).

{
  "mcpServers": {
    "actionlake": {
      "url": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer al_YOUR_KEY"
      }
    }
  }
}

Fill in a key for Kiro →

Gemini CLI

One commandAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. Run the command. -s user adds it for every project; leave it off for this project only.
2. Editing settings.json by hand? Use httpUrl. In Gemini CLI, url means the old SSE transport.
3. Signing in instead? Leave the header off and run /mcp auth actionlake in Gemini CLI.

gemini mcp add -s user --transport http --header "Authorization: Bearer al_YOUR_KEY" actionlake https://connect.actionlake.com/mcp

{
  "mcpServers": {
    "actionlake": {
      "httpUrl": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer al_YOUR_KEY"
      }
    }
  }
}

Fill in a key for Gemini CLI →

Microsoft Copilot Studio

Web formkey (X-API-Key)sign-inOAuth clientNot yet tested by us; from the vendor's docs, 2026-10-22

Generative orchestration must be on for the agent. Your tenant's data policies (DLP) govern MCP connections like any connector.

1. Open your agent, then Tools > Add a tool > New tool > Model Context Protocol.
2. Server name: ActionLake. Server URL: https://connect.actionlake.com/mcp. Write a description that says when to use it, for example "Run actions in our business systems".
3. Authentication, simplest first: OAuth 2.0 > Dynamic discovery. Copilot Studio registers itself; you sign in when you create the connection.
4. Or API key: Type Header, Header name X-API-Key. Paste your key when you create the connection.
5. Or OAuth 2.0 > Manual, with a client from Create OAuth client below. Leave its redirect URIs empty: Copilot Studio shows its callback only after Create, and an empty list accepts any https callback.
6. Click Create, then Create a new connection, then Add to agent.

Header name   X-API-Key
API key       al_YOUR_KEY

MCP server URL     https://connect.actionlake.com/mcp
Client ID          <the client ID from Create OAuth client>
Client secret      <the client secret, shown once when you create the client>
Authorization URL  https://api.actionlake.com/oauth/authorize
Token URL          https://api.actionlake.com/oauth/token
Refresh URL        https://api.actionlake.com/oauth/token
Scopes             actions account

Fill in a key for Microsoft Copilot Studio →

Microsoft 365 Copilot

Web formsign-inOAuth clientNot yet tested by us; from the vendor's docs, 2026-10-22

Microsoft 365 Copilot reaches MCP servers through a declarative agent. API keys are not supported for MCP plugins; it signs in with OAuth. The account needs custom app upload and Copilot access enabled.

1. In VS Code, install Microsoft 365 Agents Toolkit (6.12 or later).
2. Create a New Agent/App > Declarative Agent > Add an Action > Start with an MCP Server.
3. Enter the server URL https://connect.actionlake.com/mcp.
4. Authentication: OAuth (with static registration). Use a client from Create OAuth client below; its redirect URI is Teams' https://teams.microsoft.com/api/platform/v1.0/oAuthRedirect.
5. Provision, then open m365.cloud.microsoft/chat. Your agent shows with a dev suffix; the first call asks you to sign in to ActionLake.

MCP server URL     https://connect.actionlake.com/mcp
Client ID          <the client ID from Create OAuth client>
Client secret      <the client secret, shown once when you create the client>
Authorization URL  https://api.actionlake.com/oauth/authorize
Token URL          https://api.actionlake.com/oauth/token
Refresh URL        https://api.actionlake.com/oauth/token
Scopes             actions account
Redirect URI       https://teams.microsoft.com/api/platform/v1.0/oAuthRedirect

Fill in a key for Microsoft 365 Copilot →

Gemini Enterprise

Web formsign-inOAuth clientNot yet tested by us; from the vendor's docs, 2026-10-22

Gemini Enterprise takes no API keys; an admin registers an OAuth client. New connections start disabled until an admin turns them on.

1. Create an OAuth client below. Its redirect URI is https://vertexaisearch.cloud.google.com/oauth-redirect.
2. Business edition: Settings & help > Manage team > Connected apps > Add MCP Server.
3. Standard or Plus edition: Google Cloud console > Gemini Enterprise > your app > Connected data stores > New data store > Custom MCP Server.
4. Choose OAuth 2.0 and fill in the fields below. Leave PKCE on.
5. Save, enable the connection, and sign in to ActionLake when Gemini asks.

MCP server URL     https://connect.actionlake.com/mcp
Client ID          <the client ID from Create OAuth client>
Client secret      <the client secret, shown once when you create the client>
Authorization URL  https://api.actionlake.com/oauth/authorize
Token URL          https://api.actionlake.com/oauth/token
Refresh URL        https://api.actionlake.com/oauth/token
Scopes             actions account
Redirect URI       https://vertexaisearch.cloud.google.com/oauth-redirect

Fill in a key for Gemini Enterprise →

Amazon Bedrock AgentCore Gateway

Web formAPI keysign-inOAuth clientNot yet tested by us; from the vendor's docs, 2026-10-22

1. In AgentCore Identity, add an API key credential provider and store your ActionLake key in it. Create it before the target.
2. Open your gateway and add a target of type MCP server, with the endpoint https://connect.actionlake.com/mcp.
3. Outbound authentication: API key, using the provider from step 1, sent in the Authorization header with the prefix Bearer.
4. Create the target. The gateway lists ActionLake's eight tools; synchronise the target if they change.
5. Per-user sign-in instead? Use OAuth (authorization code) with a client from Create OAuth client below.

Header   Authorization
Prefix   Bearer
Key      al_YOUR_KEY

MCP server URL     https://connect.actionlake.com/mcp
Client ID          <the client ID from Create OAuth client>
Client secret      <the client secret, shown once when you create the client>
Authorization URL  https://api.actionlake.com/oauth/authorize
Token URL          https://api.actionlake.com/oauth/token
Refresh URL        https://api.actionlake.com/oauth/token
Scopes             actions account

Fill in a key for Amazon Bedrock AgentCore Gateway →

Amazon Quick

Web formsign-inNot yet tested by us; from the vendor's docs, 2026-10-22

Amazon Quick sends no custom headers, so it signs in with OAuth. It needs a Quick Enterprise subscription.

1. Open Connectors > Create for your team > Model Context Protocol (MCP).
2. Name it ActionLake and enter the endpoint https://connect.actionlake.com/mcp. Connection type: public.
3. Quick discovers ActionLake's sign-in and registers itself. Sign in to ActionLake when asked.
4. If ActionLake's tools change, open the connector and click Sync.

https://connect.actionlake.com/mcp

Fill in a key for Amazon Quick →

Salesforce Agentforce

Web formsign-inOAuth clientNot yet tested by us; from the vendor's docs, 2026-10-22

Agentforce takes no API keys for MCP servers; it uses an OAuth client you register.

1. Create an OAuth client below and leave its redirect URIs empty: Salesforce shows its callback only after you save, and an empty list accepts any https callback.
2. In Setup, open Agentforce Registry and click New (or API Catalog > MCP Servers > External Servers > Add MCP Server).
3. Name it ActionLake, describe when to use it, and enter the URL https://connect.actionlake.com/mcp.
4. Authentication: OAuth 2.0, with the client ID, client secret, endpoints and scopes below.
5. Choose the tools to allow. Then assign the permission set Salesforce created to the users or agent user who will use it; without it the tools fail.

MCP server URL     https://connect.actionlake.com/mcp
Client ID          <the client ID from Create OAuth client>
Client secret      <the client secret, shown once when you create the client>
Authorization URL  https://api.actionlake.com/oauth/authorize
Token URL          https://api.actionlake.com/oauth/token
Refresh URL        https://api.actionlake.com/oauth/token
Scopes             actions account

Fill in a key for Salesforce Agentforce →

OpenAI API and Agents SDK

CodeAPI keyNot yet tested by us; from the vendor's docs, 2026-10-22

The Responses API asks for approval before every MCP call unless you say otherwise. ActionLake's key grants and approval policy still gate what the key may do.

1. Responses API: add an mcp tool. OpenAI sends authorization as Authorization: Bearer.
2. Agents SDK: open the server with MCPServerStreamableHttp; the SDK runs the MCP client itself, with any model.
3. Give the agent a key with only the grants it needs. Make one below.

from openai import OpenAI

client = OpenAI()
response = client.responses.create(
    model="gpt-5",
    tools=[{
        "type": "mcp",
        "server_label": "actionlake",
        "server_url": "https://connect.actionlake.com/mcp",
        "authorization": "al_YOUR_KEY",
        "require_approval": "never",
    }],
    input="Use actionlake: call help, then tell me what you can do.",
)
print(response.output_text)

import asyncio
from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

async def main():
    async with MCPServerStreamableHttp(
        name="actionlake",
        params={
            "url": "https://connect.actionlake.com/mcp",
            "headers": {"Authorization": "Bearer al_YOUR_KEY"},
        },
        cache_tools_list=True,
    ) as server:
        agent = Agent(name="Assistant", instructions="Use actionlake to act on business systems.", mcp_servers=[server])
        result = await Runner.run(agent, "Call help, then tell me what you can do.")
        print(result.final_output)

asyncio.run(main())

Fill in a key for OpenAI API and Agents SDK →

REST, from your own code

CodeAPI keyVerified in ActionLake's code

1. The same eight tools answer as plain HTTPS under the same names. The OpenAPI document describes every route.
2. No key yet? One unauthenticated call creates a tenant and returns its key, shown once.
3. Every other request sends the key as Authorization: Bearer (or X-API-Key).

curl -X POST https://api.actionlake.com/v1/account/tenant.create \
     -H "Content-Type: application/json" -d '{"name": "acme"}'

curl "https://api.actionlake.com/v1/actions?q=invoice" \
     -H "Authorization: Bearer al_YOUR_KEY"

Fill in a key for REST, from your own code →

n8n

Web formAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

1. In an AI Agent node, add a tool: MCP Client Tool.
2. Endpoint: https://connect.actionlake.com/mcp. Server Transport: HTTP Streamable.
3. Authentication: Bearer Auth. Create a credential with your key as the token.
4. Tools to Include: All.

Endpoint           https://connect.actionlake.com/mcp
Server Transport   HTTP Streamable
Authentication     Bearer Auth
Bearer Token       al_YOUR_KEY

Fill in a key for n8n →

Zapier

Web formAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

Zapier's MCP Client app is in beta.

1. In a Zap, add an action from the MCP Client app.
2. Connect a new account. Server URL below; Transport: Streamable HTTP.
3. With a key: OAuth No, and paste the key as the Bearer Token. With sign-in: OAuth Yes, then sign in to ActionLake.
4. Pick the tool to run, for example do, and map its inputs.

Server URL     https://connect.actionlake.com/mcp
Transport      Streamable HTTP
OAuth          No
Bearer Token   al_YOUR_KEY

Fill in a key for Zapier →

Any other MCP client

Config fileAPI keysign-inVerified in ActionLake's code

1. Add a remote MCP server with the URL. ActionLake speaks the Streamable HTTP transport; choose it, not SSE.
2. Send the key in any one of the three headers, or leave headers off and sign in with OAuth on the same URL.
3. OAuth: PKCE (S256) with dynamic client registration. The consent page can create a free account in one click.
4. Ask your agent to call help. Eight tools: help, find_actions, get_action, do, ask, check, cost, account.

URL       https://connect.actionlake.com/mcp
Header    Authorization: Bearer al_YOUR_KEY
   or     Authorization: al_YOUR_KEY
   or     X-API-Key: al_YOUR_KEY
Sign-in   OAuth 2.1 on the same URL, instead of a header

{
  "mcpServers": {
    "actionlake": {
      "type": "http",
      "url": "https://connect.actionlake.com/mcp",
      "headers": {
        "Authorization": "Bearer al_YOUR_KEY"
      }
    }
  }
}

Fill in a key for Any other MCP client →

Stdio-only clients

Config fileAPI keysign-inNot yet tested by us; from the vendor's docs, 2026-10-22

For a client that can only start local servers, such as Claude Desktop's claude_desktop_config.json or JetBrains AI Assistant. Needs Node.js.

1. Add the server to the client's MCP config. mcp-remote runs locally and forwards to ActionLake over HTTP.
2. The header has no space after the colon on purpose: some clients split arguments on spaces. The space lives in the environment variable instead.
3. Signing in instead? Remove the two --header lines and the env block; mcp-remote opens the ActionLake consent page in your browser.
4. Restart the client completely after editing its config.

{
  "mcpServers": {
    "actionlake": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://connect.actionlake.com/mcp",
        "--transport",
        "http-only",
        "--header",
        "Authorization:${AUTH_HEADER}"
      ],
      "env": {
        "AUTH_HEADER": "Bearer al_YOUR_KEY"
      }
    }
  }
}

Fill in a key for Stdio-only clients →