v1.0.0

Highlights

Code Graph Engine: Tree-sitter based multi-language code parsing with automatic relationship graph building for functions, classes, interfaces, and dependencies.

AI-Powered Analysis: Deep code understanding with MCP (Model Context Protocol) server support, providing tools for semantic search, dependency tracking, and impact analysis for AI assistants.

Comprehensive Tooling: 20+ CLI commands for building, querying, auditing, and analyzing codebases with incremental indexing for efficient large-scale code handling.

Multi-Language Support: First-class support for Go, TypeScript, JavaScript, Python, Java, Rust, C/C++, and C# with extensible parser architecture, uses gotreesitter lib.

Visual Interface: React 19 + Sigma.js web UI with interactive graph visualization, code browsing, and AI assistant integration.

Advanced Analytics: Code complexity metrics, circular dependency detection, co-change analysis, data flow tracking, and control flow graph generation.

Features

Core Analysis Capabilities:

• Code graph building with automatic node and edge extraction
• Incremental indexing for efficient updates on large codebases
• Multiple search modes: keyword, semantic (embedding-based), regex, and hybrid
• Path finding and call chain analysis between symbols
• Impact analysis for change propagation understanding

Code Quality & Metrics:

• Circular dependency detection and reporting
• Code complexity analysis with cyclomatic complexity metrics
• Dead code identification
• Code ownership mapping from git history
• Co-change analysis to identify files that frequently change together

Advanced Analysis Tools:

• Control Flow Graph (CFG) generation for functions
• Data flow analysis to track variable usage
• Branch comparison for understanding changes across git branches
• Snapshot creation for temporal code analysis
• Diff impact analysis for pull request review
• Process detection and tracing for execution flows
• Community detection using Louvain algorithm
• PageRank for code importance ranking
• Architecture rule checking with deny rules

MCP Server Integration:

• 30+ tools for AI assistants via Model Context Protocol
• Keyword search, semantic search, and hybrid search capabilities
• Symbol lookup, callers/callees analysis, and path finding
• Impact analysis and call chain discovery
• Complexity metrics, co-change information, and PageRank
• Architecture compliance checking and community detection
• Process tracing for execution flow understanding
• Cognitive context engine (bimodal embedding + scenario templates)
• File system tools: read_file, write_file, run_command
• Agent delegation for multi-agent collaboration

Plugin System:

• Full plugin lifecycle management: install, activate, deactivate, uninstall
• Plugin manifest protocol with permissions, activation events, and platform overrides
• Frontend plugin SDK with iframe isolation, component library, and theme support
• Backend plugin support with health check, environment variables, and CORS handling
• Plugin data directory isolation per plugin with configurable uninstall modes
• Shared state and event bus for inter-plugin communication
• Lazy loading via activation events — zero resource usage when not enabled
• Official extension repository: axons-extension-packages

Web Interface:

• Interactive graph visualization using Sigma.js
• Code browser with syntax highlighting (Prism.js)
• Project management dashboard
• AI-powered chat interface
• Settings panel for embedding providers and LLM configuration
• Real-time build progress and event streaming

CLI Tools:

• build - Build code graph from source with incremental updates
• query - Query nodes, edges, and relationships
• search - Multi-mode search (keyword/semantic/regex)
• audit - Comprehensive code quality audit with cycle detection
• complexity - Analyze code complexity metrics
• owners - Map code ownership from git history
• path - Find paths between symbols
• sequence - Generate call sequence diagrams
• cfg - Generate control flow graphs
• dataflow - Analyze data flow relationships
• cochange - Identify co-changing files
• diff-impact - Analyze impact of git diffs
• branch-compare - Compare branches for differences
• snapshot - Create and manage code snapshots
• triage - Triage issues and identify affected code
• embed - Generate embeddings for semantic search
• export - Export graph data in various formats
• watch - Real-time file monitoring and graph updates
• registry - Manage multiple projects
• stats - Get comprehensive project statistics

Storage & Performance:

• SQLite-based persistent storage with project isolation
• Global database for projects, settings, and agent profiles
• Per-project databases for nodes, edges, files, and embeddings
• Embedding cache for reuse until code changes
• Lazy loading of project databases on demand

Embedding Providers:

• OpenAI (text-embedding-3-small/large)
• Ollama (nomic-embed-text, etc.)
• Jina AI embeddings
• Custom embedding endpoints
• Multi-provider support with configurable defaults

AI Agent System:

• ReAct Agent: Reasoning and Acting pattern with iterative tool execution
• Multi-Agent Orchestration: Master orchestrator delegates tasks to specialized sub-agents
• 5 Built-in Agent Profiles:
  • Orchestrator (default): Task decomposition and agent delegation
  • Architect: Module boundaries, dependency analysis, architecture compliance
  • Code Quality Analyst: Complexity, dead code, hotspots, coupling detection
  • Impact Analyst: Change impact scope, call chains, blast radius assessment
  • Code Engineer: Read/write files, execute commands, complete coding tasks
• 30+ Agent Tools: Search, graph analysis, architecture, quality metrics, file operations, command execution
• Agent Delegation: Orchestrator can delegate subtasks to specialized agents via delegate_to_agent tool
• Conversation Memory: SQLite-based session management with context persistence
• Streaming Events: Real-time thinking, tool execution, and token streaming

LLM Integration:

• OpenAI (GPT-4, GPT-3.5, etc.)
• Anthropic (Claude)
• Ollama (local models)
• Custom LLM endpoints
• Tool calling support for function execution
• Multi-round conversation with context management

Technical Details

Architecture:

• CLI and Web UI share core business logic (internal/core)
• HTTP API server for web interface with REST endpoints
• WebSocket/SSE for real-time event streaming
• MCP server for AI assistant integration
• ReAct agent with reasoning and tool execution

Requirements:

• Go 1.25+ for backend services
• Node.js 22+ for frontend development
• Tree-sitter for code parsing
• SQLite for storage

Build & Deploy:

• Cross-platform compilation (Linux, macOS, Windows)
• Docker containerization with embedded frontend
• Systemd service support
• Makefile-based build automation

Dependencies:

• github.com/modelcontextprotocol/go-sdk for MCP
• gonum.org/v1/gonum for graph algorithms
• modernc.org/sqlite for pure Go SQLite
• github.com/spf13/cobra for CLI
• github.com/odvcencio/gotreesitter for Tree-sitter bindings

Agent Skills

Three skills following the Agent Skills Specification:

Skill	Description
code-graph-analyzer	Architecture analysis, code quality metrics, and comprehensive audit reports
dependency-tracker	Dependency mapping, circular dependency detection, and impact analysis
code-search-assistant	Multi-mode search (keyword, semantic, regex, embedding) with context-aware results

All skills:

• Follow the Agent Skills standard specification
• Include YAML frontmatter with name, description, license, and metadata
• Provide CLI-based workflows with axons commands
• Can be used with any Agent Skills compatible AI assistant

Breaking Changes

Initial release - no breaking changes from previous versions.

Maintenance

Language Support:

• Go, TypeScript, JavaScript (production-ready)
• Python, Java, Rust, C/C++, C# (supported)
• Extensible architecture for adding new languages

Testing:

• Unit tests for extractors (Go, TypeScript, Python, Java, Rust)
• Integration tests for graph operations
• Test data fixtures for parser validation