Bindry

npx add-mcp 'https://api.bindry.ai/api/mcp'

MCP Server

Give your AI tool a direct connection to your Bindry workspace. It can look up your Binders and read skills when it needs them, using Model Context Protocol (MCP).

The server is hosted by Bindry, so there is nothing to run locally. Connect with a workspace API key; the available tools read content and check versions without changing your workspace.

Transport

Streamable HTTP

Authentication

Workspace API key

https://api.bindry.ai/api/mcp

1. Create a workspace API key

Open Team in the workspace you want to connect. Create an API key and copy it when it is shown. The key belongs to that workspace; it cannot access your other workspaces. Your existing Bindry plugin key works here too.

About API keys

Set the key in the environment that starts your AI tool. Replace the example value with your key. These commands apply to the current terminal session.

export BINDRY_API_TOKEN='your-workspace-api-key'

For a desktop app, make the variable available to the app's launch environment and restart it. Keep the key out of shared config files and source control.

2. Connect your AI tool

Run this in your terminal. Codex reads BINDRY_API_TOKEN from its environment and sends it as a Bearer token.

codex mcp add bindry --url https://api.bindry.ai/api/mcp --bearer-token-env-var BINDRY_API_TOKEN

Run codex mcp get bindry to check the saved configuration, then start a new Codex session with the environment variable set. Codex MCP documentation

Other MCP clients

Add a remote HTTP server with the URL above. Set the X-Api-Key header to your workspace key using your client's secret settings. Authorization: Bearer <key> also works. Your client must support HTTP MCP and custom authentication headers. Bindry's MCP connection uses API keys rather than a browser sign-in or OAuth flow.

3. Check the connection

Ask your AI tool to use Bindry to list the Binders in your workspace. A successful response, including an empty list for a new workspace, confirms the connection. A saved server entry alone does not prove the key was accepted.

Use Bindry to list the Binders in my workspace.

Connecting makes the tools available; it does not automatically apply every Binder to every conversation. Ask your agent to read the relevant skills. For compiled skills and pinned or live sync, use a Bindry plugin.

Available tools

bindry.binders.list

List the Binders in the connected workspace.

bindry.binders.get

Read a Binder, its ordered skills, and their pinned versions.

bindry.skills.get

Read a skill's current instructions, constraints, and verification checklist.

bindry.binders.check_drift

Compare local skill version pins with a Binder's current pins.

Skill reads return current content. Drift checks compare a Binder's pinned versions; those pins may differ from a skill's latest content. These tools require a workspace key, including when you have been browsing the public Library.

Troubleshooting

Unauthorized or missing API key

Check that the key is active and the environment variable is available to your AI tool. Restart the tool after changing its environment. For a revoked or expired key, create a replacement on Team.

The wrong Binders, or an empty workspace

The key selects the workspace. Create a key from the intended workspace's Team page; changing workspace in your browser does not change an existing key.

The URL does not open as a web page

That is expected. This is an MCP endpoint for HTTP POST requests, not a website. Use your client's MCP connection check and a tool call to verify access.

Your client asks for OAuth or only supports SSE

Choose HTTP transport with an API-key header. If the client cannot send that header or a Bearer token, use one of the clients above.