Interaction Bench
Interaction Bench is Interaction Bench evaluates coding agents on real interaction defects recovered from merged pull requests in real products, replayed in locked environments and judged from before/after evidence.. It is ranked #616 on the Vibe Coding Radar, in Code review & safety, first seen 7 days ago and shared in 1 post (12.4k views).
Interaction Bench - Benchmarked on what users feel. v0.1 preview v0.1 preview Benchmarked on what users feel. A feature can work correctly and still be hard for people to find, understand, or use. These difficulties can emerge as an app reaches more users and changes over time. Interaction Bench tests whether coding agents can investigate these difficulties from a user’s perspective and fix the app. See the leaderboard Why does usability matter? 44 tasks 25 open-source apps 827.5k combined GitHub stars 34.1% best pass rate [01] leaderboard Overall performance Full leaderboard Task pass rates and run statistics across the evaluated coding…
What people said about Interaction Bench on X
今天，我们 @lobehub 团队发布了一个我觉得很有意义的新Benchmark：Interaction Bench。 一直以来，我们都觉得Agent的评测和真实用户体验之间，存在着相当大的距离。 真实用户不会告诉Agent哪一行代码出了Bug，也不会给它一份详细的修复方案。用户只会说，我想实现这个功能，或者这个东西坏了，你帮我解决一下。 所以，这次我们尝试换一种方式来评估Agent。 Interaction Bench从25个真实的开源应用中提取了44个任务，只告诉Agent用户想实现的目标，让它自己探索环境、发现问题、尝试解决，并验证最终结果。 首轮结果里GPT-6 Astra以34.1%排名第一，Claude Opus…
— @MaxForAI, 7 days ago · 96 likes · see the post
Alternatives to Interaction Bench
- Introducing Vercel CI — Vercel CI is in Beta. It runs only the checks your change affects, on compute sized to the push, using the Turborepo…
- StarSling — You can also send this URL to your agent to get the CI playbooks from Linear, Anthropic, OpenAI, etc
- evilmartians.com — Rita Klubochkina turned those checks into an agent skill set and we tested it on our website. The skill ships only the…
- Weave — Weave Router is an open-source LLM router for Claude Code, Codex, and other coding agents. It routes each request to…
- Agent Baseline — Four security boundaries for software factories, where AI agents turn tickets into tested code, enforced outside the…
- cursor.com — Secure what AI writes. Find us on the Cursor marketplace
Interaction Bench in numbers
- Rank on the Vibe Coding Radar: #616 of 2591
- Shared in 1 post by 1 account: @MaxForAI
- 12.4k views on those posts
- First seen 7 days ago, last shared 7 days ago
- Pricing seen by AGK Intelligence: not stated
- Market: Code review & safety
FAQ
What is Interaction Bench?
Interaction Bench evaluates coding agents on real interaction defects recovered from merged pull requests in real products, replayed in locked environments and judged from before/after evidence. It was first shared on X 7 days ago and is ranked #616 on the Vibe Coding Radar.
Is Interaction Bench free?
Pricing is not stated on the page we read.
Who shared Interaction Bench?
1 account on X, including @MaxForAI, in 1 post totalling 12.4k views.
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 7.9k posts from 5k X accounts over the last 14 days, 2.6k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 13:28 UTC. Full method.