AIHelms MCP Gateway
AIHelms is an organizational MCP gateway that attributes calls to employee keys and tracks configured per-tool costs.
- 1,030 ★ on the repository
- Last commit
- GPL-3.0
Broad access
Why this level
- Employee keys expose operations on registered servers
- Administrative logs can retain work-request content
- The platform stores provider credentials and routes paid traffic
Install
- Works in
Manual install
git clone https://github.com/beizhu-1209/AIHelms.git
cd AIHelms
cp .env.example .envSet secrets in the local file, then run `docker compose up -d`. Add upstream MCP servers through the admin console.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Employee keys expose operations on registered servers
- Administrative logs can retain work-request content
- The platform stores provider credentials and routes paid traffic
This is third-party code. Review the repository files before installing.
What it does
The AIHelms MCP gateway gives employees access to an organization’s tools. An administrator adds a server to the catalog, refreshes its tool list and publishes it for users. Employees copy the supplied configuration into their MCP clients. A personal AIHelms key associates each call with a user and their permissions. The admin console provides MCP logs and settings for the cost of each tool call. The gateway is part of AIHelms, deployed on your own server. The platform also connects model APIs and tracks department and project budgets. The organization configures its own external-service credentials.
Who it is for. Administrators distributing corporate MCP connections and tracking employee use of shared tools.
Good fit when
- Several employees use shared MCP tools through personal keys.
- An administrator registers a server and publishes it in the shared catalog.
- Calls are attributed to users and internal costs.
- Model and tool access are managed through one administration system.
Not a fit when
- You need one local server without a corporate catalog or user accounting
- Administrators must not be able to inspect request and response content in logs
- You require bundled model subscriptions: the platform does not provide API access
Example request
Prepare a test MCP connection through AIHelms, grant it to one user and show the configuration without the key value.MCP
- Transport
- http, sse
- Authentication
- API key
Environment variables
POSTGRES_PASSWORDrequired, secret- PostgreSQL password for this deployment.
SECRET_KEYrequired, secret- Platform session-signing key.
SUPER_ADMIN_PASSWORDrequired, secret- Initial super-administrator password.
Limitations
Installation requires Docker 20.10 or later, Compose v2 and at least 4 CPU cores and 8 GB of RAM; PostgreSQL and Redis run alongside the platform. Upstream MCP permissions determine operations available through an AIHelms key; the gateway does not make them read-only. Logs may retain requests, responses and metadata, so restrict who can read them in advance. Internet access needs separate connection encryption and console sign-in permissions.
How to disable. Revoke the user key or unpublish the server in AIHelms and remove its MCP client connection. Stop the deployment with `docker compose down` in its directory. Retain or delete database and log data separately under your organization’s policy.
FAQ
How do I deploy AIHelms?
Clone the repository, copy `.env.example` to `.env` and set the secrets. Run `docker compose up -d`, then open `/admin` at your deployment address. The initial username is `admin`; the password comes from `SUPER_ADMIN_PASSWORD` in `.env`.
How does an employee obtain an MCP connection?
After the server is published, the employee opens the user portal and copies the client configuration. It specifies a `/<server-name>/mcp` address and the personal key in the `x-litellm-api-key` header.
Which server transports are supported?
SSE and Streamable HTTP. The administrator selects the transport and enters the upstream MCP address and required headers.
Who can read log contents?
`LOG_DETAIL_CONTENT_VISIBLE_TO_ADMIN`, added in version 0.1.26, controls ordinary administrators’ access. Super administrators can read requests, responses and metadata regardless of this setting.
How does the gateway differ from an MCP server?
It manages server connections and access. Registered servers still perform database, file and external-service operations.
What license does AIHelms use?
The repository uses GPL-3.0. The author also describes a commercial licensing option; choose the applicable terms before distributing a modified build.
Discussion
No questions or reviews yet.
Related
MCP serversA self-hosted gateway routes internal MCP tools, authenticates users and records requests for later review.
A CLI runs a model and MCP-tool gateway with owned API keys, format translation, routing and request logs.
A self-hosted gateway that merges several MCP servers into one endpoint with auth and tool selection
A self-hosted MCP and A2A gateway with identity provider integration, access control and observability for enterprise agents
A Node.js manager for MCP servers: it starts local servers, connects remote ones and gives a client one address, with a REST API for control
MCP Gateway
mcp-gateway
A reverse proxy and lifecycle management for MCP servers on Kubernetes and Azure from Microsoft: routing, authorization, roles. For platform teams
README in short
The employee portal opens at the deployment’s main address. HTTP API documentation is available at `/api/docs`.