Contextual

The product is a local context engine that provides temporal semantic search over code repositories, commits, pull requests and documentation to retrieve relevant functions, decisions and history that AI assistants otherwise forget. It is for software developers and engineering teams in businesses that need to query and understand evolving codebases. It is delivered as a single on-device binary that indexes and stores embeddings locally with zero cloud egress and connects to editors and agents via Model Context Protocol.

Key features

• Temporal semantic search
• Plain English code search
• Temporal queries over history
• Code graph with calls and imports
• Blame-enriched history
• Impact analysis and blast radius
• Undeclared coupling detection
• Architectural decision records
• Hybrid search via MCP
• Filesystem watcher auto-reindex
• Branch-aware timeline forking
• On-device embeddings with CodeRankEmbed
• Local LanceDB index storage