Code intelligence platform. Understand, search, and review code at depth -- powered by an MCP server for AI agents, a VS Code extension with Copilot Chat integration, a Web Dashboard, and a standalone CLI.
# Install and setup in one command
curl -fsSL https://raw.githubusercontent.com/AgentiX-E/code-analyzer/main/scripts/setup.sh | bash
# Or via npm
npm install -g @code-analyzer/cli
# Initialize a project and index it
cd your-project
code-analyzer init
code-analyzer analyze .
# Search your codebase
code-analyzer search "authentication"
# Review code for issues
code-analyzer review src/
# Auto-detect and configure your AI agent
code-analyzer agent detect
code-analyzer agent configure| Capability | Description |
|---|---|
| Knowledge Graph | Builds a rich property graph of your codebase -- 33 node types, 40+ relationship types with in-memory and SQLite storage |
| 18-Phase Pipeline | DAG-based analysis pipeline: scan -> parse -> scope resolution -> communities -> embeddings |
| PR Review | 5-stage review pipeline with 50+ rules across 6 categories (security w/ CWE, correctness, performance, maintainability, style, architecture). Smart file bundling for PR review context, delegation review mode |
| Cross-Repo Analysis | Multi-repo indexing, federated search, API contract detection, version matrix, cross-repo PR review with GitHub check runs. Incremental reindexing with git change detection |
| 20-Language Support | TypeScript, JavaScript, Python, Go, Java, Kotlin, C#, Rust, Ruby, PHP, Swift, C/C++, C-like + IaC: Dockerfile, HCL/Terraform, YAML, JSON, Markdown, Shell, SQL |
| Cypher Query Engine | Full Cypher query support with 40+ graph relationship types for deep structural queries across the knowledge graph |
| MCP Server | 40 tools, 15 resources, 5 prompts for AI agents -- auth, sliding-window rate limiter, Cypher queries |
| VS Code Extension | Copilot Chat participant with 15 slash commands: /review, /explain, /impact, /find, /deps, /refactor, /test, /analyze, /coverage, /standards, /review-deps, /check-contract, /trace-dataflow, /find-hotspots, /audit-security |
| Web Dashboard | 6 interactive views: Graph Explorer, Search, Dashboard, Cross-Repo, PR Review Panel, Repo Group Manager |
| GitHub Integration | Webhook receiver, cross-repo PR review bridge, check runs with annotations, repo sync, REST + GraphQL API client |
| AI Agent Integrations | Auto-detection and one-click setup for 11 agents: Claude Code, Cursor, Windsurf, Continue.dev, Aider, Cline, GitHub Copilot, Codex, Gemini CLI, Cody, Amazon Q |
| Enterprise Security | RBAC (5 roles/25 permissions), audit logging, 16-pattern secret scanner |
| Operational Excellence | Health checks, graceful shutdown, retry with exponential backoff, dead letter queue, sliding-window rate limiter. Auto-index on project open, auto-watch file changes, comment positioning with precision validation, graph artifact compression and sharing |
+---------------------------------------------------------------+
| Presentation Layer |
| CLI (6 commands) | VS Code Extension (Copilot) |
| Web Dashboard (6 views) |
+---------------------------------------------------------------+
| Integration Layer |
| MCP Server (40 tools) | HTTP API | Webhook (PR) |
+---------------------------------------------------------------+
| Service Layer |
| Review Pipeline | Search Engine | Standards |
+---------------------------------------------------------------+
| Intelligence Layer |
| 50+ Rules (CWE) | Cross-Repo | Impact | Embeddings |
+---------------------------------------------------------------+
| Analysis Engine |
| 20 Lang (Tree-sitter + IaC) | 18-Phase DAG | Graph Builder |
+---------------------------------------------------------------+
| Infrastructure Layer |
| Graph Store | Worker Pool | Git Ops | Parallel |
+---------------------------------------------------------------+
| Foundation Layer |
| Config | Logging | Errors | i18n | Metrics |
+---------------------------------------------------------------+
## Benchmarks
Code Analyzer achieves **industry-leading results** with zero LLM token cost:
| Metric | Code Analyzer | SonarQube AI | Augment Code | CodeRabbit | GitHub Copilot |
|--------|:---:|:---:|:---:|:---:|:---:|
| **Precision** | **79.4%** | 72% | 65% | 58% | 42% |
| **Recall** | **73.0%** | 48% | 55% | 52% | 38% |
| **F1 Score** | **0.761** | 0.576 | 0.596 | 0.549 | 0.399 |
| **Noise Rate** | **0.3x** | 0.8x | 1.5x | 2.1x | 3.2x |
| **Cost** | **$0** | API cost | API cost | API cost | API cost |
[Full benchmark report →](docs/BENCHMARK_REPORT.md)
> **Note**: Benchmarks are based on internal test suites. Independent third-party validation is planned for future releases.
| Package | Description |
|---|---|
@code-analyzer/shared |
Shared types (33 node labels, 39 edges), constants, protocols |
@code-analyzer/core |
Foundation: config, logging, errors, i18n, metrics, agent detection, security, RBAC, audit |
@code-analyzer/infra |
Infrastructure: graph stores, file discovery, git operations, worker pool |
@code-analyzer/analyzer |
Analysis: 20-language tree-sitter + IaC, 18-phase pipeline, scope resolution, auto-index, auto-watch, incremental reindexing |
@code-analyzer/intelligence |
Intelligence: 50+ rules, cross-repo, impact analysis, embeddings, standards, GitHub client |
@code-analyzer/mcp |
MCP server: 39 tools, 15 resources, 5 prompts, Cypher engine, middleware |
@code-analyzer/server |
HTTP REST API server with webhook support, rate limiting, graceful shutdown |
@code-analyzer/cli |
Full CLI: init, analyze, search, review, status, agent |
@code-analyzer/vscode |
VS Code extension with Copilot Chat participant (7 slash commands) |
@code-analyzer/web |
Web Dashboard with 6 interactive views |
- Getting Started -- Installation, quick start, first analysis
- Architecture -- 7-layer design, data flow, design decisions
- MCP Server -- Tool reference, resources, prompts, configuration
- Code Review & PR Review -- Rules reference, PR review workflow, standards, review swarm
- Web Dashboard -- Interactive UI guide, hooks, API client
- GitHub Integration -- Webhooks, cross-repo PR review, check runs, repo sync
- Language Support -- 12-language tree-sitter coverage
- Configuration -- Options, environment variables, tuning
- Integrations -- AI agent setup guides
- API Reference -- REST API specification, endpoints, examples
- Troubleshooting -- Common issues, performance tuning, debugging
# Install dependencies
pnpm install
# Run all tests (5,200+ tests, 95%+ coverage across all dimensions)
pnpm test
# Build all packages
pnpm build
# Run a specific package's tests
pnpm --filter @code-analyzer/cli test
# Lint and typecheck
pnpm lint
pnpm typecheck# Docker (multi-arch: amd64, arm64)
docker compose up -d
# Homebrew (macOS)
brew install code-analyzer
# Manual build from source
git clone https://github.com/AgentiX-E/code-analyzer.git
cd code-analyzer
pnpm install && pnpm build
node packages/mcp/dist/index.jsMIT (c) Lambertyan