stanford-iris-lab/meta-harness-tbench2-artifact
Agent scaffold for AI benchmarking on Terminal-Bench 2.0 that injects sandbox environment snapshots into prompts.
Not currently ranked — collecting fresh signals.
star history
Meta-Harness is an agent development scaffold built on Terminus-KIRA and Harbor’s Terminus-2 framework, designed for evaluating AI agents on terminal task benchmarks. It improves agent performance by pre-injecting environment context (working directory, tools, packages) into the initial prompt, eliminating early exploration turns. The system achieves 76.4% accuracy across 89 tasks using Claude Opus 4.6, with the optimal agent configuration discovered through automated harness evolution.
Frequently asked
- What is stanford-iris-lab/meta-harness-tbench2-artifact?
- Agent scaffold for AI benchmarking on Terminal-Bench 2.0 that injects sandbox environment snapshots into prompts.
- Is meta-harness-tbench2-artifact open source?
- Yes — stanford-iris-lab/meta-harness-tbench2-artifact is an open-source project tracked on heatdrop.
- What language is meta-harness-tbench2-artifact written in?
- stanford-iris-lab/meta-harness-tbench2-artifact is primarily written in Python.
- How popular is meta-harness-tbench2-artifact?
- stanford-iris-lab/meta-harness-tbench2-artifact has 1.1k stars on GitHub.
- Where can I find meta-harness-tbench2-artifact?
- stanford-iris-lab/meta-harness-tbench2-artifact is on GitHub at https://github.com/stanford-iris-lab/meta-harness-tbench2-artifact.