A framework for recursive optimization of AI agent harnesses (prompts, tools, memory, skills) without fine-tuning model weights, with JobBench, LAB and WorkBuddy-Bench benchmarks.
Useful for
- Run the local harness2-demo without model credentials
- Run the LAB benchmark with parallel recursion and evaluate results
- Compare base and refined harnesses on JobBench or WorkBuddy-Bench
Before using
- License: Apache-2.0
- Installation and requirements have not been verified; review the README before running
Why now
- Created on GitHub
- 2026-10-06 UTC
- First seen by GitNova
- 2026-10-18 UTC
- Typical day
- +1 stars · median of 11 complete days
- Magnitude
- 8.9 / 10 · growing interest
Already 1573.0× a typical complete day; a spike in interest, not a quality score
The current gain already exceeds the maximum of the previous 11 complete days: +1,329 on 2026-10-17 (UTC)
Measured: October 18, 2026, 12:37 UTC
Measurement details
- Created on GitHub: 2026-10-06; first seen by GitNova: 2026-10-18 (UTC)
- +1,573 stars measured 2026-10-18, 02:46–12:26 UTC
- Usually: +1 stars/day — median of 11 complete days 2026-10-07–2026-10-17 (UTC)
- Already 1573.0× a typical complete day; a spike in interest, not a quality score
- The current gain already exceeds the maximum of the previous 11 complete days: +1,329 on 2026-10-17 (UTC)
- Magnitude 8.9/10