How many bugs can LMs find & fix in large codebases?
[NeurIPS 2026] FedAgent: A Library for Decentralized Agent Learning. 🏆 Best Paper Award at the AAAI'26 TrustAgent Workshop, 🏆 Outstanding Paper Award in the AAAI'26 PerFM Workshop.
Baba Is Harbor - AI agent benchmark on Baba is You
An end-to-end framework for Agentic Reinforcement Learning.
Adversarial probes for Harbor benchmark integrity. Ocarina Labs' Harbor extension — null-agent, output-echo, judge-injection, and verifier-tamper probes that flag broken benchmarks before they ship.
A broken Terminal-Bench 2 (Harbor) task, repaired: reproducible env, solution-leak removed, gameable verifier replaced with real assertions, instructions aligned 1:1 with the verifier.
Run agent evaluations with Harbor in Google Kubernetes Engine at scale
Playground for reproducible Harbor eval tasks that test local coding agents with offline, Docker-backed verifiers.
Terminal-Bench evals using your own Pi agent
To associate your repository with the harbor-framework topic, visit your repo's landing page and select "manage topics."