@zhelunsun

ai-research-harness enforces cognitive discipline for llm-agent experimentation through repo scaffolding, six design disciplines, and seven governance rules. covers experiment phases from setup through handoff, with emphasis on reproducibility, evidence grading, and preventing overclaiming before scaling.

skillfactory codifies a four-phase sop (capture, collect, distill, publish) for turning raw ideas into versioned, distributable openclaw skills. includes structured upgrade workflow with semantic versioning and clawhub publishing automation.

skill covers agent-first engineering principles from openai's harness work, with emphasis on codebase structure, context management, and observability for autonomous agents. lacks actionable procedures and concrete failure handling.

Provides file-based protocols to enable audit trails, shared context, mission control, flight recording, and seamless handoff in multi-agent collaboration.

Cognitive discipline for AI-native scientific experimentation. Trigger when setting up controlled experiments with LLM agents, designing reproducible evaluat...