GitHub · Python
KrystalUnity/krystal-loop-protocol
A low-cost model does the coding, a frontier model checks the exact result. 59 of 66 benchmark tasks solved at $0.046 each, against 56 at $0.721 for the heavy model alone. One-command runner with receipts you can recompute.
- Stars
- 14
- Forks
- 0
- Stars gained in 24 hours
- 0
- Forks gained in 24 hours
- 0
Topics: agent-orchestration, agentic-coding, ai-agents, ai-coding, benchmark, capability-preservation, code-review, coding-agents, cost-optimization, developer-tools
Latest data point: Nov 20, 2026 · 14 stars · 0 forks