Patch the Planet: global response coordination as agents uncover critical vulnerabilities

OpenCode is an open source AI coding agent that's become a popular alternative to Claude Code and Cursor. But hackers got into part of its online platform and stole some users’ API keys, and the AI tools it turned to for help kept refusing to look at the malicious code.

Hi HN, we’re Robert and Din from Laminar ( https://laminar.sh ). We’ve been building open-source observability for AI agents since 2024. Recently, we released flow-1, a model trained with RL to find errors in agent traces. On our benchmark ( https://laminar.sh/blog/flow-1 ), i...

Existing benchmarks measure solve rate and token output/cost, but don't punish excess code, which makes a codebase harder to work with in the long term. I wanted to measure this since it matters most for long-term projects. Method: it takes the patches that benchmarks already ...