ContextLoom helps developers search, understand, and navigate complex codebases using explainable AI-powered hybrid retrieval. Local-first. Zero cloud dependency.
Combine exact token matching with semantic understanding. Get precise results without losing context.
Your code stays on your machine. Embeddings generated locally via Ollama. No telemetry, no cloud sync.
See why a result ranked highly. Transparent scoring: vector similarity + keyword match + symbol boost.
Isolated indices per repository. Switch projects instantly without reindexing or context loss.
Query RepoHandler and get the class definition — not unrelated property getters. Our hybrid
approach balances lexical exactness with semantic understanding.
Every result includes a score breakdown. Understand how vector similarity, keyword matches, and symbol boosts contributed to ranking.
.repo-awareness.yaml lets you tune thresholds, weights, and skip patterns per repository. One platform, many workflows.
Session isolation, content-hash caching, graceful fallbacks, and audit-ready logging. Built for real-world development teams.
ContextLoom is currently in active research and development.