@garrytan: Interesting thing GBrain can do now: If you have a skill + code + test + resolver + resolver trigger + evals you want t…
Summary
GBrain can now package skills, code, tests, and evals into a SKILLPACK tarball that others can install using a simple command.
View Cached Full Text
Cached at: 05/20/26, 12:29 PM
Interesting thing GBrain can do now: If you have a skill + code + test + resolver + resolver trigger + evals you want to package for someone else to use…
GBrain will package it up for you into what I call a SKILLPACK
It’s tarball and anyone else can install it using the same GBrain skillpack command
Similar Articles
@garrytan: GBrain v0.36 just dropped. Brand new README, and totally redone skillpack system that makes each bundle of skills, code…
GBrain v0.36 is released with a new README and a revamped skillpack system that allows users to customize and update skill bundles independently.
@garrytan: GBrain SkillOpt now has 4 E2E evals that verify it working https://github.com/garrytan/gbrain-evals/blob/main/docs/benc…
Garry Tan's gbrain-evals is an open-source test suite for gbrain, an AI agent's long-term memory, with 4 end-to-end evaluations verifying SkillOpt functionality, achieving high recall and precision on multiple benchmarks.
@garrytan: This is interesting. Anyone experimenting with this? So far anytime I have adjacent skills I just tell it to DRY itself…
Garry Tan asks if others experiment with merging adjacent AI skills into larger parameterized skills, sharing his preference for composing bigger skills with branching parameters.
@garrytan: GBrain v0.36.1 just landed on main. It ships something that was an idea from a hackathon team on Saturday: a team creat…
GBrain v0.36.1 adds full evals inspired by a hackathon team's Hindsight concept for tracking predictions over time.
@Voxyz_ai: can't wait for this gbrain feature. here's the loop: agent attempts a task using a skill ↓ gbrain eval or LLM-as-judge …
Voxyz announces a new GBrain feature that enables agents to iteratively improve skills using LLM-as-judge evaluation and an overnight optimization cycle.