OpenHands

For coding agents, "skills" are a great way to automate repetitive workflows, but how can we tell if they're working at scale? We did a deep dive on how you can log, monitor, and improve agent skills, with a real example of building a customized PR review skill.

paper: arxiv.org/abs/2603.17829 code: github.com/OpenHands/code… models/data: huggingface.co/collections/Op…

Check out our new research project with OpenHands community members and CMU on training the strongest open-source language model for agentic code search!

Skill quality is task-dependent and model-dependent. Don't assume a skill helps just because it exists. Tutorial repo: github.com/rajshah4/evalu… Blog: openhands.dev/blog/evaluatin…

Task 3: sales pivot analysis. Overall pass rate went from 70% to 80% with the skill. But the effect varied by model. Some improved. Some regressed. The skill nudged one model into a brittle path that made it less reliable. Skills can be counterproductive. You have to measure.

Skills are becoming a core building block for AI coding agents. But some skills make the agent worse. We ran three tasks across five models to show how to measure when skills actually help - and when they don't.

Ever wonder what happens when software teams move from copilots to autonomous agents? Next week in #Boston, we’re teaming up with OpenHands to unpack what this shift actually looks like in practice—from how work gets done across the SDLC to how engineering leaders measure real impact. Join Robert Brennan (CEO, OpenHands) and Nick Arcolano (Head of Research, Jellyfish) for a candid conversation on what’s working, what’s not, and what’s changing fast. 🕠 Tuesday, March 24 @ 5:30 PM ET 📍 Pillar VC, Boston ⚠️ Nearly full, RSVP here: luma.com/ai-meetup-24ma…

Yikes. A lot of “skills” actually make agents worse. We assume adding a skill improves performance. In reality, it often introduces new failure modes, increases confusion, and can lower pass rates on real tasks. The tricky part is that this isn’t always obvious. A skill might work in a demo, feel smarter, and still make the agent less reliable overall. So the real question is: How do you know if a skill is actually helping?

Velocity is dead. If AI can generate a compiler, “write more code faster” isn’t the constraint anymore. Our Chief Architect, Ray Myers, on what matters next: reliability, constraints, and agent systems. openhands.dev/blog/20260219-…

Check out how you can profile and improve agent skills using @OpenHandsDev and @lmnrai here: openhands.dev/blog/20260227-…

Congrats to the Laminar team on their raise! It has been excellent working with them so far on profiling and improving agent skills. More and more love for the open agent toolstack ❤️

We're excited to be featured in Jensen's keynote at @nvidia GTC as one of the AI Native leaders in AI for Software Development! Looking forward to continuing to build a strong, robust ecosystem for open-source AI!

This Saturday, March 14, AI builders will gather in San Francisco for the Arena. During the Opening Day, we’ll be joined by speakers from OpenHands (@openhandsdev), alphaXiv (@askalphaxiv), Dedalus (@dedaluslabs), Daytona (@daytonaio), and Sentient to dig into grounded reasoning and what it takes to make agents reliable in real-world scenarios. We’ll close with time to meet Cohort 0 and get early ideas on the table before the sprint begins.

#AI coding tools are everywhere — but which #LLM actually fits YOUR workflow? 🤔 @VMblog sat down with Graham Neubig of @OpenHandsDev, to talk benchmarking, open source vs. closed models, and how to pick the right AI for your dev team. vmblog.com/qa/benchmarkin… #SoftwareEngineering

Want to see where OpenHands is headed next? 👀 Join our call TODAY. We will be presenting our roadmap and want feedback from YOU. RSVP below 👇️

New model release by @nvidia - Nemotron 3 Super! x.com/ctnzr/status/2… We got early access to test it in OpenHands and it works well, excited to have a great new locally deployable LLM.

@pillar_vc @_jellyfish_co luma.com/ai-meetup-24ma…

Are you located in #Boston? Interesting in learning more about #AIAgents? Join our event at @pillar_vc with @_jellyfish_co