SACL MCP Server

SACL MCP Server is an MCP server that implements the SACL research framework to deliver bias-aware code retrieval and code localization for AI coding assistants. It connects tools like Claude Code and Cursor to a local Neo4j or Graphiti knowledge graph and OpenAI embeddings. Developers, software engineers, and AI code agents use SACL to overcome textual bias in software repositories, where traditional search over-relies on docstrings, comments, and variable names rather than underlying behavior. The server analyzes codebases to extract functional signatures, structural features, and dependency call graphs. When an AI client issues a search, SACL detects potential bias indicators, recalculates similarity using semantic-augmented vectors, and localizes the exact code blocks matching functional intent. It also provides tools to inspect relationships, monitor bias metrics, and incrementally update changed files, ensuring coding agents retrieve functionally correct code even when documentation is sparse or misleading.

Category: AI & LLM Tooling

Tags: bias-detection, code analysis, code-retrieval, reranking, semantic search

Visit SACL MCP Server

How to install and configure SACL MCP Server

1. Ensure Node.js 18+, a running Neo4j database, and an OpenAI API key are available. 2. Clone the repository and install dependencies: bash git clone https://github.com/ulasbilgen/sacl cd sacl npm install 3. Copy the sample environment file and set your credentials: bash cp .env.example .env 4. Build the application: bash npm run build 5. In your claude_desktop_config.json, register the server under mcpServers: json { "mcpServers": { "sacl": { "command": "node", "args": ["/path/to/sacl/dist/index.js"], "env": { "OPENAI_API_KEY": "your-key", "NEO4J_URI": "bolt://localhost:7687", "NEO4J_USER": "neo4j", "NEO4J_PASSWORD": "password" } } } }

What you can do with SACL MCP Server

• Search codebases for specific computational patterns without relying on exact variable names, comments, or docstrings. - Identify and score textual bias levels in source files across docstrings, identifiers, and comments. - Map file dependencies, import-export structures, and call graphs to provide full contextual components to AI assistants. - Execute incremental updates on single or batched files to maintain knowledge graph accuracy during active editing sessions. - Locate exact code segments responsible for specific algorithmic functions during AI-assisted refactoring tasks.

Key facts

• https://github.com/ulasbilgen/sacl
• AI & LLM Tooling, Developer Tools & Code Intelligence
• bias-detection, code analysis, code-retrieval, reranking, semantic search

Part of MCP Servers

What is SACL MCP Server?

SACL MCP Server is a Model Context Protocol service providing bias-aware code retrieval. It detects when code search relies too heavily on surface text like comments or variable names and reranks results using semantic signatures and graph-based code dependencies.

Which MCP clients work with SACL MCP Server?

SACL MCP Server works with any MCP-compatible environment, including Claude Code, Claude Desktop, and Cursor IDE, as well as custom AI agents that implement the Model Context Protocol.

What external services does SACL MCP Server require?

The server requires an active Neo4j database instance for semantic knowledge graph storage and an OpenAI API key for generating code embeddings and completing language model analysis.

What tools are provided by SACL MCP Server?

It provides tools including analyze_repository, query_code, query_code_with_context, update_file, update_files, get_relationships, get_file_context, get_bias_analysis, and get_system_stats.