Vera: a Rust code indexing tool for AI agents combining BM25, embeddings and reranking
lemon07r · reddit · 2026-11-11
After nearly a year of work, a developer rebuilt his code indexing/search tool Vera from scratch in Rust (open source, 127 stars), designed for AI coding agents. His lesson from running a coding eval leaderboard: context and token efficiency are everything.
- Pain point: agents waste tokens guessing grep terms in unfamiliar codebases. Vera does semantic search over what code does, returning ranked functions with paths, line numbers and symbols, plus structural search and reference lookup.
- Design: returns compact Markdown code blocks (no LLM-generated explanations), Tree-sitter parsing preserves boundaries, and an accompanying skill teaches agents when to use semantic search vs rg vs not searching. Recommended integration is CLI + skill; MCP also supported.
- Retrieval stack: BM25 + embeddings + rank fusion + code-aware ranking, fully local, no external vector DB. Recommended setups: fully local with MinishLab Potion Code v2 (CPU-friendly, offline); API with Qwen3 Embedding 8B + Qwen3 Reranker 8B for best retrieval quality.
- Validated generalization on separate repos and 40 agent sessions on cross-file questions, avoiding benchmark overfitting.
More from coding & agent
- Natural language search over autonomous driving data with Cosmos-Embed1 + OpenSearch — 4310sy · 2026-11-11
- Spawning an 'interpretability researcher' subagent via a clever system prompt — Sauers_ · 2026-11-11
- Building a walkable 3D home from floor plans with Claude — full prompts shared — FinanceYF5 · 2026-11-11
- Two prompts run an entire AI-powered 3D real-estate tour business, from prospecting to delivery — FinanceYF5 · 2026-11-11
- Tektronix launches ~$500 12-bit DPO scope with AI MCP support in closed beta — IgorCarron · 2026-11-11
- Agent Bridge: Open-Source MCP Server Drives Your Logged-in Chrome, Cutting Token Use 57% vs Claude in Chrome — falconsw · 2026-11-11