Skip to content

Roadmap

CARI (Code-Aware Retrieval Index) is production-ready with 92 tests passing across 5 implementation phases. The Knowledge Graph pipeline is functional but considered optional/advanced.

  • Improved scoring heuristics for ranked retrieval
  • Better handling of monorepo structures
  • Support for more annotation source types
  • Python AST extraction (tree-sitter)
  • Go AST extraction (tree-sitter)
  • Rust AST extraction (tree-sitter)
  • Generic fallback for unsupported languages (regex-based symbol detection)
  • iw index watch — continuous re-indexing on file changes
  • Richer report output with actionable suggestions
  • Better error messages and onboarding flow
  • Import/dependency graph integration into connections
  • Test coverage correlation (if test framework metadata available)
  • PR review history as a co-change signal
  • Comment/TODO tracking
  • VS Code extension with inline drift warnings
  • Hover providers showing CARI context for symbols
  • CodeLens for annotation density
  • Pre-built GitHub Action (uses: intentweave/check-drift@v1)
  • Support for GitLab CI, CircleCI templates
  • Baseline comparison (track drift trend over time)

The KG pipeline (LLM extraction → Neo4j) is functional but under evaluation:

  • Useful for architecture exploration and decision archaeology
  • Cost (LLM API calls) and complexity (Neo4j) limit adoption
  • Keeping it as an optional advanced path
  • May evolve based on community feedback and use cases

Have ideas? See Contributing or open a GitHub Discussion.