MCP connector
One click. Every agent connected.
The ahel MCP connector is a remote MCP server your agents read over one link. Pick your agent below; the click does the setup.
Works with Claude, Claude Code, Codex, Cursor, VS Code, Copilot, Windsurf, and any MCP client, all at https://mcp.ahel.ai.
Connect
Pick your agent.
Open Claude’s connector screen
The click also carries your link along, so the modal is a single drop.
Add to ClaudeName it ahel
In the add-connector modal, name the connector ahel and drop the address the click brought with it.
Sign in and ask away
OAuth once in the browser. From then on Claude reads your gateway on every conversation.
The protocol
How does MCP work?
The Model Context Protocol gives agents one standard way to call tools. A connector is an address that answers those calls. Most connectors serve one product; this one serves your account, so the same link means something different for every person who signs in: their gateway, their servers, their credentials.
The tools
What the link serves today
Every client that connects sees these six tools, plus the hosted MCP servers on your gateway, served through this same link today with each tool namespaced under its server key. Enabled skills are served too, as MCP prompts and through skill_read. Local and stdio servers are the next leg of the build.
| Tool | What it does |
|---|---|
| stack_list | What is on your gateway and which items the link is serving right now. |
| catalog_search | Search the whole catalog of MCP servers, skills, subagents, and plugins from inside your agent. |
| stack_add | Add a catalog item to your gateway. It lands available, not served, until you enable it. |
| stack_set_enabled | Turn one item on or off. On means served through the link; off keeps it on the map, unserved. |
| credentials_status | Which credential types are sealed in your store and which are missing. Status only: values never leave. |
| skill_read | The full instructions of any enabled skill on your gateway. Enabled skills also arrive as MCP prompts, which Claude Code surfaces as slash commands. |
The boundary
What it deliberately does not do
Your agent can
- Search the whole catalog from inside a conversation.
- Add servers to your gateway and switch them on or off; that only ever changes your own link.
- Call the tools of every server you serve, credentials attached server-side.
It never can
- Read, write, or reveal a stored credential. No tool on this link touches a key, and a credential never crosses back to the client.
- Reach another account. The link answers only as the person who signed in.
- Seal a new credential. That stays in the app, at ahel.ai.
Questions
Everything else.
What is MCP?
The Model Context Protocol is the open standard AI agents use to reach tools and data. An MCP connector is an address an agent reads capabilities from. ahel's connector serves your whole account at one address.
How does the ahel MCP connector work?
It is a remote MCP server over streamable HTTP. Your agent calls the link, OAuth signs you in once, and every call after that is answered with your gateway: no local install, no config to maintain.
Which agents does this work with?
Claude, Claude Code, Cursor, Codex, VS Code, Copilot, Windsurf, and any client that speaks MCP over streamable HTTP. The link is the same one for all of them.
Does it work with ChatGPT?
Any ChatGPT surface that supports MCP connectors can use the same link. Where connector support has not shipped yet, it works the moment it does: the link does not change.
What can my agent actually reach?
Only what your account can already reach: your gateway, the catalog, and the tools of the hosted MCP servers you have enabled. Credential values never cross the link. Nothing more, and nothing from anyone else's account.
Can it spend money?
The link's own tools cannot: they only read the catalog and edit your gateway. But a call proxied to a server you enabled does whatever that server does, with your credential attached, so a paid vendor call spends money the moment your agent makes it.
Connect it once.
Every agent you use reads the same tools from then on.