codeprism

CodePrism is an MCP server implemented in Rust that provides a production-ready set of code intelligence tools via the Model Context Protocol. The server exposes a suite of 20 MCP tools organized into Core Navigation, Search & Discovery, Analysis, and Workflow Orchestration, with multi-language support for JavaScript/TypeScript and Python. After building the server, start the binary to listen for JSON-RPC 2.0 requests from AI assistants or clients. You can interact with the server to perform repository analysis, symbol discovery, dependency tracing, pattern and security analyses, and real-time updates to code graphs. The Mandrel MCP Test Harness is included for validating protocol compliance and running test specifications against the server, ensuring stable, repeatable behavior in automated environments.

Prerequisites:

• Rust 1.82+ (stable toolchain)
• Git
• A compatible repository to analyze (JavaScript/TypeScript or Python)

Installation steps:

1. Clone the repository: git clone https://github.com/rustic-ai/codeprism cd codeprism

2. Build the MCP server in release mode: cargo build --release

3. Run the MCP server (example): ./target/release/codeprism

4. Verify the server startup and available commands: ./target/release/codeprism --help

Notes:

• The server reads REPOSITORY_PATH and other environment variables at runtime as needed by plugins and analyzers.
• For production deployments, consider wrapping the binary in a process manager (systemd, PM2 if using Node tooling conventions, or similar) and configure logging via RUST_LOG.

Tips and common considerations:

• Ensure Rust toolchain (1.82+) is installed and up to date.
• The Mandrel MCP Test Harness can validate YAML-based MCP tests against the server. Install via cargo if you need to run tests locally.
• Set REPOSITORY_PATH to the root of the workspace you intend to analyze so CodePrism can auto-detect repository structure.
• If you encounter port or transport issues, check firewall rules and ensure the JSON-RPC endpoint is reachable by the client or AI assistant.
• Environment variables can tune logging, performance, and repository detection behavior; consult docs/MANDREL_PROJECT_OVERVIEW.md for advanced options.