XERJ open-sourced: index similar GitHub repos before coding, cuts AI tokens by ~62% vs grep
lxfater · x · 2026-10-18
lxfater shares the tool he always wished existed: XERJ (8.2k GitHub stars) downloads related open-source repos, auto-indexes them into searchable data, and lets coding agents consult proven implementations before writing code instead of grepping raw sources.
- In a controlled test on an unfamiliar codebase, both groups passed 16/16 runs, but XERJ users generated 62% fewer tokens than the grep group.
- The project claims 40x token savings over grep and positions itself as a durable ElasticSearch replacement for search, RAG, security audits and agent memory.
More from coding & agent
- Developer: depending on two unreliable AI labs is chaos — open weights win long-term — StewartalsopIII · 2026-10-18
- Codex surprise: it opened its own ChatGPT chat to send itself a task — LaunchAxis · 2026-10-18
- Epic ships official plugin letting Claude Code control Unreal Editor via MCP — CSProfKGD · 2026-10-18
- Indie team starts building Revit replacement Mutiny just 24 hours after trigger email — anselm · 2026-10-18
- Visualizing 10K agent trajectories: LLM-labeled treemaps expose classifier disagreements — enjalot · 2026-10-18
- 'Cuck coding': the savage new nickname for just watching AI write your code — petesena · 2026-10-18