{
  "markdown": "[![CI](https://github.com/m1rl0k/Context-Engine/actions/workflows/ci.yml/badge.svg)](https://github.com/m1rl0k/Context-Engine/actions/workflows/ci.yml)\n[![npm version](https://img.shields.io/npm/v/@context-engine-bridge/context-engine-mcp-bridge.svg)](https://www.npmjs.com/package/@context-engine-bridge/context-engine-mcp-bridge)\n[![VS Code Marketplace](https://img.shields.io/visual-studio-marketplace/i/context-engine.context-engine-uploader.svg?label=VS%20Code)](https://marketplace.visualstudio.com/items?itemName=context-engine.context-engine-uploader)\n[![Join our Discord](https://img.shields.io/badge/Discord-Join%20Chat-5865F2?logo=discord&logoColor=white)](https://discord.gg/tCxvqmP4QT)\n\n**Documentation:** [Getting Started](docs/GETTING_STARTED.md) · README · [Configuration](docs/CONFIGURATION.md) · [IDE Clients](docs/IDE_CLIENTS.md) · [MCP API](docs/MCP_API.md) · [ctx CLI](docs/CTX_CLI.md) · [Memory Guide](docs/MEMORY_GUIDE.md) · [Architecture](docs/ARCHITECTURE.md) · [Multi-Repo](docs/MULTI_REPO_COLLECTIONS.md) · [Observability](docs/OBSERVABILITY.md) · [Kubernetes](deploy/kubernetes/README.md) · [VS Code Extension](docs/vscode-extension.md) · [Troubleshooting](docs/TROUBLESHOOTING.md) · [Development](docs/DEVELOPMENT.md)\n\n---\n\n## Context-Engine\n\nOpen-source, self-improving code search that gets smarter every time you use it.\n\n<p align=\"center\">\n  <img src=\"useage.png\" alt=\"Context-Engine Usage\" width=\"50%\"/>\n</p>\n\n---\n\n## Quick Start: Stack in 30 Seconds\n\n### VS Code Extension (Easiest)\n1. Install [Context Engine Uploader](https://marketplace.visualstudio.com/items?itemName=context-engine.context-engine-uploader)\n2. Open any project → extension prompts to set up Context-Engine stack\n3. Opened workspace is indexed\n4. MCP configs can configure your agent/IDE\n\n**That's it!** The extension handles everything:\n- Clones Context-Engine to your chosen location (keeps it separate from your project)\n- Starts the Docker stack automatically\n- Sets up MCP bridge configuration\n- Writes MCP configs for Claude Code, Windsurf, and Augment\n\n**Claude Code users:** Install the skill plugin:\n```\n/plugin marketplace add m1rl0k/Context-Engine\n/plugin install context-engine\n```\n\n### Manual Setup (Alternative)\n```bash\ngit clone https://github.com/m1rl0k/Context-Engine.git && cd Context-Engine\nmake bootstrap  # One-shot: up → wait → index → warm → health\n```\n\nOr step-by-step:\n```bash\ndocker compose up -d\nHOST_INDEX_PATH=/path/to/your/project docker compose run --rm indexer\n```\n\n*See [Configuration](docs/CONFIGURATION.md) for environment variables and [IDE_CLIENTS.md](docs/IDE_CLIENTS.md) for MCP setup.*\n\n---\n\n## Why This Stack Works Better\n\n| Problem | Context-Engine Solution |\n|---------|------------------------|\n| **Large file chunks** → returns entire files | **Precise spans**: Returns 5-50 line chunks, not whole files |\n| **Lost context** → missing relevant code | **Hybrid search**: Semantic + lexical + cross-encoder reranking |\n| **Cloud dependency** → vendor lock-in | **Local stack**: Docker Compose on your machine |\n| **Static knowledge** → never improves | **Adaptive learning**: Gets smarter with every use |\n| **Tool limits** → only works in specific IDEs | **MCP native**: Works with any MCP-compatible tool |\n\n---\n\n## What You Get Out of the Box\n\n- **ReFRAG-inspired micro-chunking**: Research-grade precision retrieval\n- **Self-hosted stack**: No cloud dependency, no vendor lock-in\n- **Universal compatibility**: Claude Code, Windsurf, Cursor, Cline, etc.\n- **Auto-syncing**: Extension watches for changes and re-indexes automatically\n- **Memory system**: Store team knowledge alongside your code\n- **Optional LLM features**: Local decoder (llama.cpp), cloud integration (GLM, MiniMax), adaptive rerank learning\n\n### Works With Your Local Files\nNo complicated path setup - Context-Engine automatically handles the mapping between your local files and the search index.\n\n### Enterprise-Ready Features\n- **Built-in authentication** with session management (optional)\n- **Unified MCP endpoint** that combines indexer and memory services\n- **Automatic collection injection** for workspace-aware queries\n\n**Alternative: Direct HTTP endpoints**\n```json\n{\n  \"mcpServers\": {\n    \"qdrant-indexer\": { \"url\": \"http://localhost:8003/mcp\" },\n    \"memory\": { \"url\": \"http://localhost:8002/mcp\" }\n  }\n}\n```\n\n*Using other IDEs? See [docs/IDE_CLIENTS.md](docs/IDE_CLIENTS.md) for complete MCP configuration examples.*\n\n---\n\n## Supported Clients\n\n| Client | Transport |\n|--------|-----------|\n| Claude Code | SSE / RMCP |\n| Cursor | SSE / RMCP |\n| Windsurf | SSE / RMCP |\n| Cline | SSE / RMCP |\n| Roo | SSE / RMCP |\n| Augment | SSE |\n| Codex | RMCP |\n| Copilot | RMCP |\n| AmpCode | RMCP |\n| Kiro | RMCP |\n| Antigravity | RMCP |\n| Zed | SSE (via mcp-remote) |\n\n---\n\n## Endpoints\n\n| Service | URL |\n|---------|-----|\n| Indexer MCP (SSE) | `http://localhost:8001/sse` |\n| Indexer MCP (RMCP) | `http://localhost:8003/mcp` |\n| Memory MCP (SSE) | `http://localhost:8000/sse` |\n| Memory MCP (RMCP) | `http://localhost:8002/mcp` |\n| Qdrant | `http://localhost:6333` |\n| Upload Service | `http://localhost:8004` |\n\n---\n\n## VS Code Extension\n\n[Context Engine Uploader](https://marketplace.visualstudio.com/items?itemName=context-engine.context-engine-uploader) provides:\n\n- **One-click upload** — Sync workspace to Context-Engine\n- **Auto-sync** — Watch for changes and re-index automatically\n- **Prompt+ button** — Enhance prompts with code context before sending\n- **MCP auto-config** — Writes Claude/Windsurf MCP configs\n\nSee [docs/vscode-extension.md](docs/vscode-extension.md) for full documentation.\n\n---\n\n## MCP Tools\n\n**Search** (Indexer MCP):\n- `repo_search` — Hybrid code search with filters\n- `context_search` — Blend code + memory results\n- `context_answer` — LLM-generated answers with citations\n- `search_tests_for`, `search_config_for`, `search_callers_for`\n\n**Memory** (Memory MCP):\n- `store` — Save knowledge with metadata\n- `find` — Retrieve stored memories\n\n**Indexing**:\n- `qdrant_index_root` — Index the workspace\n- `qdrant_status` — Check collection health\n- `qdrant_prune` — Remove stale entries\n\nSee [docs/MCP_API.md](docs/MCP_API.md) for complete API reference.\n\n---\n\n## Documentation\n\n| Guide | Description |\n|-------|-------------|\n| [Getting Started](docs/GETTING_STARTED.md) | VS Code + dev-remote walkthrough |\n| [IDE Clients](docs/IDE_CLIENTS.md) | Config examples for all supported clients |\n| [Configuration](docs/CONFIGURATION.md) | Environment variables reference |\n| [MCP API](docs/MCP_API.md) | Full tool documentation |\n| [Architecture](docs/ARCHITECTURE.md) | System design |\n| [Multi-Repo](docs/MULTI_REPO_COLLECTIONS.md) | Multiple repositories in one collection |\n| [Kubernetes](deploy/kubernetes/README.md) | Production deployment |\n\n---\n\n## How It Works\n\n```mermaid\nflowchart LR\n  subgraph Your Machine\n    A[IDE / AI Tool]\n    V[VS Code Extension]\n  end\n  subgraph Docker\n    U[Upload Service]\n    I[Indexer MCP]\n    M[Memory MCP]\n    Q[(Qdrant)]\n    L[[LLM Decoder]]\n    W[[Learning Worker]]\n  end\n  V -->|sync| U\n  U --> I\n  A -->|MCP| I\n  A -->|MCP| M\n  I --> Q\n  M --> Q\n  I -.-> L\n  I -.-> W\n  W -.-> Q\n```\n\n---\n\n## Language Support\n\nPython, TypeScript/JavaScript, Go, Java, Rust, C#, PHP, Shell, Terraform, YAML, PowerShell\n\n---\n\n## Benchmarks\n\n### CoSQA (Dense Retrieval, No Rerank)\n\n| Method | MRR | R@1 | R@5 | R@10 | NDCG@10 |\n|--------|-----|-----|-----|------|---------|\n| **Context-Engine (Jina-Code)** | **0.276** | 0.146 | 0.448 | 0.658 | 0.365 |\n| Context-Engine (BGE-base) | 0.253 | 0.150 | 0.374 | 0.550 | 0.322 |\n| CodeT5+ embedding | 0.266 | - | - | - | - |\n| BM25 (Lucene) | 0.167 | - | - | - | - |\n| BoW | 0.065 | - | - | - | - |\n\n*Corpus: 20,604 code snippets | 500 queries | Pure dense retrieval, no reranking*\n*Jina-Code: jinaai/jina-embeddings-v2-base-code (code-specific, 8k context)*\n\n---\n\n## License\n\nMIT\n\n",
  "bytes": 7902,
  "sha": "6e096e259cef021d3c439e3369e5ba33fc2acedb43680e4ac4c6da2cfcdd5409",
  "repo_slug": "m1rl0k/context-engine",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/skl_m1rl0k_context_engine_skills_context_eng_6f4fb5a8/readme"
}