CodePrism
STDIOAI-generated MCP server providing graph-based code intelligence and analysis with 23 production-ready tools
AI-generated MCP server providing graph-based code intelligence and analysis with 23 production-ready tools
⚠️ IMPORTANT: This project is entirely AI-generated. Not a single byte of code, documentation, or configuration has been written by humans. This is an experimental project showcasing the capabilities of AI-driven software development.
A production-ready, high-performance code intelligence server implementing the Model Context Protocol (MCP). CodePrism provides AI assistants with structured understanding of codebases through graph-based analysis, enabling real-time, accurate code intelligence.
This project represents a unique experiment in software development:
Want to contribute? See our Contributing Guidelines for exciting ways to participate without writing code!
✅ 20 Production-Ready Tools - 100% success rate, no failed tools
✅ Full MCP Compliance - JSON-RPC 2.0 with complete protocol implementation
✅ Multi-Language Support - JavaScript/TypeScript + Python with advanced analysis
✅ Semantic APIs - User-friendly parameter names, no cryptic IDs required
✅ Environment Integration - Automatic repository detection via REPOSITORY_PATH
✅ Parser Development Tools - Complete debugging and development toolkit
CodePrism is proudly sponsored by Dragonscale Industries Inc, pioneers in AI innovation and development tools.
Dragonscale Industries Inc supports the development of cutting-edge AI-powered code intelligence, enabling CodePrism to remain open-source and freely available to the developer community. Their commitment to advancing AI technology makes projects like CodePrism possible.
Become a sponsor → | Learn more about sponsorship →
┌─────────────────┐ MCP Protocol ┌──────────────────┐
│ AI Assistant │◄──────────────────►│ codeprism-mcp-server │
│ (Claude/Cursor)│ JSON-RPC 2.0 │ Server │
└─────────────────┘ └──────────────────┘
│
┌────────────┴────────────┐
┌───────────────▼───────────────▼─────────────────┐
│ 20 MCP Tools │
│ ┌─────────────┐ ┌─────────────────────────┐ │
│ │ Core │ │ Search & Discovery │ │
│ │ Navigation │ │ 4 tools │ │
│ │ 4 tools │ └─────────────────────────┘ │
│ └─────────────┘ ┌─────────────────────────┐ │
│ ┌─────────────┐ │ Analysis │ │
│ │ Workflow │ │ 11 tools │ │
│ │ 4 tools │ │ │ │
│ └─────────────┘ └─────────────────────────┘ │
└─────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────┐
│ Graph-Based Code Analysis │
│ JavaScript/TypeScript + Python Support │
└─────────────────────────────────────────────────┘
NEW: CodePrism now includes the Mandrel MCP Test Harness - a comprehensive testing framework for MCP servers built on the official Rust SDK.
# Install and run moth binary cargo install --path crates/mandrel-mcp-th # Test MCP servers with YAML specifications moth test filesystem-server.yaml # Validate test specifications moth validate filesystem-server.yaml
# Clone and build git clone https://github.com/rustic-ai/codeprism cd codeprism cargo build --release # Verify installation ./target/release/codeprism --help
⚠️ Development Note: This project enforces strict implementation completeness standards via git pre-commit hooks. All commits must contain complete, functional implementations with zero placeholder code. The existing .git/hooks/pre-commit script automatically validates code quality and implementation completeness.
📝 Note on Repository Setup: The server starts without a specific repository. Once connected via MCP, use any analysis tool (like
repository_stats) and the server will prompt you to specify the repository path, then automatically initialize and index it.
🏆 Claude Desktop - Best overall MCP experience
// ~/.config/claude-desktop/claude_desktop_config.json { "mcpServers": { "codeprism": { "command": "/path/to/codeprism/target/release/codeprism", "args": ["--mcp"], "env": { "CODEPRISM_PROFILE": "development", "RUST_LOG": "info" } } } }
⚡ Cursor - AI pair programming with code intelligence
// .cursor/mcp.json { "mcpServers": { "codeprism": { "command": "/path/to/codeprism/target/release/codeprism", "args": ["--mcp"], "env": { "CODEPRISM_PROFILE": "development", "RUST_LOG": "info" } } } }
🔧 Manual Usage - Direct stdio communication
# Set configuration and run export CODEPRISM_PROFILE=development export RUST_LOG=info ./target/release/codeprism --mcp
repository_stats - Get comprehensive repository overview and statisticsexplain_symbol - Detailed symbol analysis with context (accepts semantic names like "UserManager")trace_path - Find execution paths between code elementsfind_dependencies - Analyze what a symbol or file depends onsearch_symbols - Advanced symbol search with regex and inheritance filteringsearch_content - Full-text search across all repository contentfind_files - File discovery with glob and regex pattern supportcontent_stats - Detailed content and complexity statisticsanalyze_complexity - Code complexity metrics and maintainability analysistrace_data_flow - Forward and backward data flow analysisanalyze_transitive_dependencies - Complete dependency chains with cycle detectiondetect_patterns - Architectural and design pattern recognitiontrace_inheritance - Python inheritance hierarchy with metaclass analysisanalyze_decorators - Python decorator analysis with framework detectionfind_unused_code - Detect unused functions, variables, and imports with confidence scoringanalyze_security - Security vulnerability detection with CVSS scoring and OWASP mappinganalyze_performance - Performance analysis with time complexity and memory usage detectionanalyze_api_surface - API surface analysis with versioning compliance and breaking change detectionfind_duplicates - Code duplication detection with similarity scoring and refactoring recommendationssuggest_analysis_workflow - Intelligent analysis guidance for specific goalsbatch_analysis - Parallel execution of multiple tools with result aggregationoptimize_workflow - Workflow optimization based on usage patternsfind_references - Complete reference analysis across the codebase# Get repository overview {"name": "repository_stats", "arguments": {}} # Analyze specific symbol {"name": "explain_symbol", "arguments": {"symbol": "UserManager"}} # Search for patterns {"name": "search_symbols", "arguments": {"pattern": "^Agent.*", "symbol_type": "class"}}
# Trace inheritance hierarchies {"name": "trace_inheritance", "arguments": {"class_name": "Agent", "include_metaclasses": true}} # Analyze decorator usage {"name": "analyze_decorators", "arguments": {"decorator_pattern": "@app.route"}} # Detect metaprogramming patterns {"name": "detect_patterns", "arguments": {"pattern_types": ["metaprogramming_patterns"]}}
# Get analysis recommendations {"name": "suggest_analysis_workflow", "arguments": {"goal": "understand_architecture"}} # Run multiple tools in parallel {"name": "batch_analysis", "arguments": {"tools": ["repository_stats", "content_stats", "detect_patterns"]}}
CodePrism is developed and maintained by Dragonscale Industries Inc, our primary sponsor and pioneer in AI innovation. Join them in supporting this project:
Your support helps us:
Become a sponsor → | View all sponsors →
👩💻 "Analyze the authentication system in this codebase"
🤖 AI uses CodePrism to:
1. Find auth-related symbols with search_symbols
2. Trace inheritance hierarchies for auth classes
3. Analyze decorator patterns for security
4. Map data flow through authentication functions
5. Provide comprehensive security analysis
👨💻 "What are the main design patterns in this Python project?"
🤖 AI leverages CodePrism to:
1. Run detect_patterns for architectural analysis
2. Use trace_inheritance for class hierarchies
3. Analyze decorators for framework patterns
4. Generate detailed architecture documentation
🔧 "Help me understand the impact of changing this class"
🤖 AI uses CodePrism to:
1. Find all references with find_references
2. Analyze transitive dependencies
3. Trace inheritance impact on subclasses
4. Assess complexity before/after changes
Benchmarked Performance:
Test Coverage:
Since this is a 100% AI-generated project, we welcome contributions in unique ways:
Remember: No code contributions accepted - but your ideas, feedback, and support drive the AI's development decisions!
CodePrism uses fully automated releases via GitHub Actions:
Via Cargo (Recommended):
cargo install codeprism-mcp-server
Download Binary:
# Linux x86_64 wget https://github.com/rustic-ai/codeprism/releases/latest/download/codeprism-linux-x86_64 chmod +x codeprism-linux-x86_64 # macOS wget https://github.com/rustic-ai/codeprism/releases/latest/download/codeprism-macos-x86_64 # Windows # Download from: https://github.com/rustic-ai/codeprism/releases/latest/download/codeprism-windows-x86_64.exe
Docker:
docker pull ghcr.io/rustic-ai/codeprism:latest docker run -e CODEPRISM_PROFILE=development -e RUST_LOG=info -v /path/to/repo:/workspace ghcr.io/rustic-ai/codeprism:latest
Our AI developer has some quirks:
Dual-licensed under MIT and Apache 2.0. See LICENSE-MIT and LICENSE-APACHE for details.
Ready to explore the future of AI-generated development tools?
⭐ Star the project to support AI-driven open source!
🐛 Report issues to help the AI improve!
💬 Join discussions to shape the AI's roadmap!
🎉 Share your experience with 100% AI-generated tooling!
"When AI writes better code than humans, it's not replacing developers—it's becoming one." - CodePrism AI Developer, 2024