Create LLM graders and run evals in JavaScript with one file
Hi HN! Run it: OPENROUTER_API_KEY="sk" npx bff-eval --demo We built a tool to help people take LLM outputs and easily grade them / eval them to know how good an assistant response is. We've built a number of LLM apps, and while we could ship decent tech demos, we were disappointed with how they'd perform over time. We worked with a few companies who had the same problem, and found out scientifically building prompts and evals is far from a solved problem... writing these things feels more like directing a play than coding. Inspired by Anthropic's constitutional ai concepts, and amazing…
In plain words
bff-eval is a JavaScript tool that lets developers create LLM graders and run evaluations from a single file. It helps users assess the quality of AI assistant responses by grading outputs against specified criteria. The tool addresses performance degradation in LLM applications over time by enabling scientific prompt building and evaluation through structured debugging techniques, inspired by constitutional AI concepts. Users can test multiple models against datasets and use actual metrics to improve response quality without retraining models.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hi HN! Run it: OPENROUTER_API_KEY="sk" npx bff-eval --demo We built a tool to help people take LLM outputs and easily grade them / eval them to know how good an assistant response is. We've built a number of LLM apps, and while we could ship decent tech demos, we were disappointed with how they'd perform over time. We worked with a few companies who had the same problem, and found out scientifically building prompts and evals is far from a solved problem... writing these things feels more like directing a play than coding. Inspired by Anthropic's constitutional ai concepts, and amazing software like DSPy, we're setting out to make fine tuning prompts, not models, the default approach to improving quality using actual metrics and structured debugging techniques. Our approach is pretty simple: you feed it a JSONL file with inputs and outputs, pick the models you want to test against (via OpenRouter), and then use an LLM-as-grader file in JS that figures out how well your outputs match the original queries. If you're starting from scratch, we've found TDD is a great approach to prompt creation... start by asking an LLM to generate synthetic data, then you be the first judge creating scores, then create a grader and continue to refine it till its scores match your ground truth scores. If you’re building LLM apps and care about reliability, I hope this will be useful! Would love any feedback. The team and I are lurking here all day and happy to chat. Or hit me up directly on Whatsapp: +1 (646) 670-1291 We have a lot bigger plans long-term, but we wanted to start with this simple (and hopefully useful!) tool. Run it: OPENROUTER_API_KEY="sk" npx bff-eval --demo
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 27d ago · cactuscompute.com
Launched alongside, June 2025
the whole month →- IM
Hey HN! Pilots everywhere are required to keep a logbook of all their flying hours, aircraft, airports, and so on. Since I track everything digitally (some people still just use paper logbooks!), I put together some data visualizations and a few 3D globes to show my flying history. This globe is probably my favourite so far: https://jameshard.ing/pilot/globes/all If you’ve got ideas for other graphs or ways to show this kind of data, I’d love to hear them!
Life & fun · 2025 · jameshard.ing
Stunning presentations with AI. No design skills required.
AI · 2025 · chroniclehq.com
Build native mobile apps for iOS and Android without code.
Dev tools · 2025 · bble.io
All your connections, across all your socials.
AI · 2025 · pally.com
- WC
I was the main contributor to workout.lol, an open-source fitness app to easily build a workout routine. The project had traction (1.4k GitHub stars, 95 forks, ~20K visits/month), but was eventually sold due to video licensing hurdles. The new owner stopped maintaining it, and the repo went abandoned. Over the next 9 months, I sent 15 emails to try to save it : no replies. Feature requests & issues were ignored. The community was left with a "broken" tool let's say. I couldn't just let it die So I built the new version from scratch with the same open-source spirit, but a better…
Dev tools · 2025 · github.com