Our take
Activeregular commits and releases- Activity
- 85
- Community
- 56
- Issues
- 83
- Pull requests
- 75
+38 stars in the last 30 days
- Releases more often than 87% of the projects we track
- Merges more pull requests than 77% of the projects we track
SocratiCode is a focused code-intelligence layer for teams that want AI assistants to navigate large repositories with searchable structure, and its active releases and contributor base indicate ongoing development.
Good fit if
- Your AI coding workflow needs more than text search, especially dependency and symbol-level analysis.
- You want a local-first MCP service that can be used across supported assistants and editor integrations.
- You can run the Node.js server and, for the default setup, Docker-managed Qdrant and Ollama.
Look elsewhere if
- You need a hosted team index immediately: the README describes SocratiCode Cloud as a private beta.
- Your environment cannot run Docker and you do not want to configure external Qdrant or another embedding setup.
- AGPL-3.0 obligations do not suit your distribution or deployment model.
All health signals
| Last commit | 2026-11-11 (2 days ago) |
|---|---|
| Commits, last 90 days | 100+ |
| Releases, last 12 months | 52 (latest v1.16.2, 2026-11-05) |
| Contributors | 24 (top contributor: 59% of commits) |
| Issues closed, last 90 days | 37+ (typically closed in 1 day) |
| Pull requests merged, last 90 days | 59 (typically merged in 0 days) |
| Project age | 9 months |
Checked on 2026-11-14 with the GitHub API.
Project history
Daily GitHub snapshots since Oct 13, 2026. History covers stars and forks; health signals are not tracked historically.
Stars
3,376Forks
428Change since Nov 15, 2026: +1 stars·0 forks
Overview
SocratiCode is a TypeScript codebase context engine built on Qdrant, the vector database used to index and search project code. It gives AI assistants a shared view of large codebases through MCP, native agent plugins, and editor integrations, addressing the limits of asking an assistant to explore files one at a time.
Its focus is automated, private code understanding: hybrid semantic and keyword search, dependency relationships, symbol-level impact and call-flow analysis, and searchable project artifacts such as schemas and API specifications. The default setup runs locally; cloud embeddings and external Qdrant are options. Supported integrations include Claude Code, OpenAI Codex, VS Code, Cursor, Gemini CLI, Continue, Cline, Zed, and OpenCode.
Key Features
- Hybrid search combines semantic vectors and BM25 keyword results, fused with Reciprocal Rank Fusion.
- AST-aware chunking organizes supported source files around functions and classes, with fallback chunking for other files.
- Dependency graphs cover more than 19 languages and support circular-dependency detection and visualization.
- Symbol tools trace callers, callees, impact, and call flows to help assess changes.
- Incremental, checkpointed indexing resumes after interruptions and can update automatically as files change.
- Cross-project search, branch-aware indexes, and shared indexes support multi-repository and multi-agent work.
- Context artifacts make database schemas, API specifications, infrastructure configurations, and architecture documents searchable alongside code.
Use Cases
- A developer joining a large or unfamiliar repository can search for a concept, then use graph and symbol tools to locate relevant code before reading files.
- A team planning a rename or refactor can inspect symbol impact and file dependencies before changing code.
- Engineers working across related services can search linked projects and trace dependencies between codebases.
- Teams using multiple AI assistants can share one indexed view rather than relying on assistant-specific code indexes.
- Organizations with private or air-gapped code can use the default local stack without sending code to a cloud embedding provider.
What you need
Detected in the repository
- Node.js >=18.17.0 (from package.json)
- A Compose file to run it with its related services
- A test suite and automated checks on GitHub Actions
License in plain words
AGPL-3.0network copyleft
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the license notice
- Share your changes: yes, the full source, also when people only use it over a network (e.g. as a hosted service)
- Includes an explicit patent grant from the contributors:
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
The README recommends Node.js 18.17 or newer and Docker for the default local setup. Add this MCP server configuration to a compatible host, then ask your assistant to index the project:
{
"mcpServers": {
"socraticode": {
"command": "npx",
"args": ["-y", "--prefer-online", "socraticode@latest"]
}
}
}
See the README for host-specific installation, configuration, and plugin instructions.
Alternatives
- codebase-memory-mcp: It focuses on persistent code-structure graphs and architectural queries, rather than combining graph analysis with semantic and keyword search.
- aft: It emphasizes Rust and tree-sitter symbol navigation, code edits, and background tasks, rather than a TypeScript Qdrant-based context engine.
- gitnexus: It builds a LadybugDB and Tree-sitter knowledge graph for MCP and browser exploration, rather than using Qdrant for hybrid search.
- Understand-Anything: It models code, documentation, and knowledge bases as interactive graphs, rather than centering on searchable code context and editor integrations.
| Project | Language | License | Stars | Status |
|---|---|---|---|---|
| SocratiCode | TypeScript | AGPL-3.0 | 3.4k | Active |
| codebase-memory-mcp | C | MIT | 45.7k | Active |
| aft | Rust | MIT | 333 | Active |
| gitnexus | TypeScript | – | 175 | Active |
| Understand-Anything | TypeScript | MIT | 85.2k | Active |
Considerations
- The default local stack relies on Docker for Qdrant and, unless a native Ollama instance is detected, embeddings. Docker is not required with external Qdrant and a suitable external embedding setup.
- Initial indexing downloads images and an embedding model in the default setup; large repositories may need more time and resources. Docker containers on macOS and Windows do not get GPU access, so the README recommends native Ollama or cloud embeddings for better indexing performance on larger projects.
- The project is actively maintained, with frequent commits and releases and contributions from multiple developers. Its development is not described as concentrated around one person, but its relative youth means teams should validate language coverage and indexing behavior against their own repositories.
- The repository is licensed under AGPL-3.0. Review its terms for your intended use, particularly if you modify or distribute the software or offer it as a service.
Found this useful?
Share it with someone who would like SocratiCode.