TrendingRisingGemsRadarToolboxCategoriesAbout

Agent · alanhuangyoo/crux

alanhuangyoo/crux

Taking the pi coding agent to Claude Code-level performance: 0.539 → 0.773 pass@1 on Terminal-Bench 2.1 with the same self-hosted 27B model, by re-engineering the agent.

Open source AI & agents TypeScriptMIT

Stars

186

of the repo behind it

Today

+0

last 24 hours

This week

+0

Forks

36

Star growth

2026-10-19: +0 (today so far)2026-10-19todaymax +1/day
Stars gained per day. Tracking since 2026-10-19; the lighter bar is today so far.
#agent-evaluation#ai-agent#coding-agent#context-engineering#llm#terminal-bench
Created
2026-08-24
Last commit
2026-10-07
First seen
2026-10-18
Gained since tracking
+0

Similar in AI & agents

Updated Mon, 19 Oct 2026 05:00:16 GMT · description from the project itself