v1.0.2
What is implemented in CostAffective MCP (v1.0.2)
Core Retrieval Engine
- Architecture-aware repository analysis
- Multi-strategy code retrieval
- Semantic repository search
- Grep-based search
- Full-text search (SQLite FTS)
- Naive fallback retrieval
- Context ranking and relevance scoring
Repository Understanding
- Project architecture detection
- Repository structure analysis
- Language-aware code discovery
- Tree-sitter based source code parsing
- Symbol and reference extraction
Context Optimization
- Context compression
- Response compression
- Context building and aggregation
- Token usage reduction strategies
- Cost-focused retrieval pipeline
Memory Systems
- Repository memory
- Discovery memory
- Knowledge memory
- Session management
Storage Layer
- SQLite-backed storage
- Persistent indexing
- Cache management
- Repository state tracking
MCP Server
- MCP-compatible stdio server
- Tool registration and execution
- Local-first architecture
- Zero-cloud operation
Installer System
- Claude Code integration
- Cursor integration
- OpenCode integration
- Codex CLI integration
- Antigravity integration
- Automatic configuration generation
- Doctor/health-check commands
Developer Experience
- Version command
- Build metadata injection
- Installation verification
- Environment diagnostics
- Configuration validation
Quality & Release Infrastructure
- GitHub Actions CI/CD
- Automated testing
- Automated linting
- GoReleaser integration
- Versioned releases
- Linux amd64 release builds
- Windows amd64 release builds
- SHA256 checksum generation
Current Focus
CostAffective MCP is designed to help coding assistants work on large repositories with lower token usage and lower cost by retrieving only the most relevant code, architecture information, and repository context instead of sending large amounts of source code to the model.