ProDisco

Progressive Disclosure Kuberentes MCP Server

typescript

2

GitHub Stars

typescript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "harche-prodisco": {
      "command": "npx",
      "args": [
        "-y",
        "@prodisco/mcp-server",
        "--config",
        "prodisco.kubernetes.yaml"
      ],
      "env": {
        "LOKI_URL": "http://localhost:3100",
        "KUBECONFIG": "${HOME}/.kube/config",
        "PROMETHEUS_URL": "http://localhost:9090",
        "PRODISCO_CONFIG_PATH": "/path/to/prodisco.config.yaml"
      }
    }
  }
}

ProDisco is a progressive-disclosure MCP server framework that indexes APIs from TypeScript libraries and runs sandboxed code against a curated, surface-level API set. This enables you to explore library capabilities safely, discover the right tools, and execute code in isolation with clean, final results returned to you.

How to use

You use ProDisco by configuring an MCP server that exposes a scoped API surface built from your chosen libraries. With a client, you discover available methods, types, and functions, write code that uses only those APIs, execute it in a sandbox, and receive the final output. Use this approach to build reproducible experiments, perform targeted automation, and maintain strong isolation between the execution environment and your host.

How to install

Follow these concrete steps to set up ProDisco MCP servers locally for Kubernetes-oriented examples and PostgreSQL in-memory testing.

Additional configuration and usage notes

Environment variables and development helpers are described below. You can run the MCP server in standard I/O mode or switch to HTTP transport for remote access. You can add multiple library configurations to index and expose their APIs for discovery.

Development setup and example runs

# Kubernetes + Observability example: start the MCP server via npm-driven workflow
curl -O https://raw.githubusercontent.com/harche/ProDisco/main/examples/prodisco.kubernetes.yaml
claude mcp add ProDisco --env KUBECONFIG="${HOME}/.kube/config" -- npx -y @prodisco/mcp-server --config prodisco.kubernetes.yaml

# PostgreSQL (in-memory testing): start the MCP server with a YAML config
curl -O https://raw.githubusercontent.com/harche/ProDisco/main/examples/prodisco.postgres.yaml
claude mcp add ProDisco -- npx -y @prodisco/mcp-server --config prodisco.postgres.yaml

# Remove when you're done
claude mcp remove ProDisco

Environment variables you may use

Configure optional environment variables before starting the MCP server. The --env flag may not reliably pass variables to the MCP server process.

If you run in a kind cluster, you can port-forward Prometheus and point the MCP server to the local URL.

HTTP transport and remote access

ProDisco can run with HTTP transport for network-based MCP connections. Start in HTTP mode on the default port or a custom port, then use the provided endpoints to initialize sessions and send tool requests.

Advanced deployment and security

For stronger isolation, you can deploy the sandbox server in a Kubernetes cluster and connect it to the MCP server over TCP. You can also enable TLS or mTLS for encrypted communications in production deployments.

Testing

Run integration tests in a KIND cluster and with the Claude Agent SDK to verify end-to-end behavior.

Notes and further reading

You will find more details about analytics, search capabilities, and sandbox architecture in the dedicated sections describing advanced analytics and sandbox architecture.

Available tools

prodisco.searchTools

Search and browse API documentation for the configured libraries, including methods, types, and functions extracted from library declaration files.

prodisco.runSandbox

Execute TypeScript code in a sandboxed environment using the same library allowlist as the search tool.