CodeCortex Context Engine

Open-source context intelligence infrastructure for AI coding agents.

CodeCortex turns a software repository into a query-specific evidence system. It combines repository structure, symbols and references, dependencies, Git/PR intelligence, impact analysis, architecture, memory, structural search, and context budgeting behind one agent-facing surface.

Retrieve evidence before generating confidence.

Start in one minute

python -m pip install --upgrade codecortex-context-engine
cortex init .
cortex index
cortex doctor

Then inspect the repository:

cortex architecture
cortex semantic "authentication and session lifecycle"
cortex impact AuthService

Or expose it over MCP:

cortex mcp --path .

See the full Getting Started guide for supported coding-agent configuration and the deterministic demo.

What it provides

• Repository maps, multi-language symbols, dependencies, and call relationships
• Precision definition/reference/implementation evidence when an index is available
• Hybrid lexical, semantic, structural, graph, Git, and memory retrieval
• Context ranking, deduplication, slicing, budgeting, and provenance
• Git, PR, impact, ownership, architecture, and drift intelligence
• Project/team memory and multi-repository workspace search
• MCP integration and merge-safe configuration for supported coding agents
• Guarded semantic editing and structural rewrite preview/apply flows
• Remote/distributed operation, platform API, SDKs, and observability
• Reproducible quality, security, benchmark, SBOM, signing, and release evidence

Navigate the docs

• Getting Started
• Architecture
• Integrations
• Evidence Fusion
• Advanced Intelligence
• Distributed operation
• Platform
• Quality
• Testing
• Licensing
• Release

The GitHub repository remains the source of truth for code, issues, security policy, releases, and contribution workflow.