Learn Harness Engineering
Learn Harness Engineering 是一门专注于 AI 编程智能体工程化落地的课程。本课程深度研究并总结了业内最前沿的 Harness Engineering(工具马具/脚手架工程)理论与实践,参考资料包括:
- OpenAI: Harness engineering: leveraging Codex in an agent-first world
- [Anthropic: Effective harnesses for long-running…
课时
- 🆕 What's New — August 2026
- 🆕 What's New — July 2026
- Syllabus
- Core References
- 文档
- zh
- lectures
- lecture-01-why-capable-agents-still-fail
- lecture-02-what-a-harness-actually-is
- code
- lecture-03-why-the-repository-must-become-the-system-of-record
- code
- lecture-04-why-one-giant-instruction-file-fails
- code
- lecture-05-why-long-running-tasks-lose-continuity
- lecture-06-why-initialization-needs-its-own-phase
- code
- lecture-07-why-agents-overreach-and-under-finish
- code
- lecture-08-why-feature-lists-are-harness-primitives
- code
- lecture-09-why-agents-declare-victory-too-early
- code
- lecture-10-why-end-to-end-testing-changes-results
- lecture-11-why-observability-belongs-inside-the-harness
- lecture-12-why-every-session-must-leave-a-clean-state
- code
- lecture-13-loop-engineering
- lecture-14-graph-engineering
- 欢迎来到项目实战
- 中文资料库
- OpenAI 高级资源包
- reference
- 模板使用指南
- Skills(技能集)
- lectures
- zh
- 动手项目
- Project 01: Baseline vs Minimal Harness
- Project 02: Agent-Readable Workspace
- Project 03: Multi-Session Continuity with Scope Control
- Project 04: Runtime Feedback and Structural Control
- Project 05: Evaluator Loops and Three-Role Upgrades
- Project 05: Evaluator Loops and Three-Role Upgrades
- solution
- starter
- Project 06: Runtime Observability and Debugging (Capstone)
- Project 06: Runtime Observability and Debugging (Capstone)
- solution
- starter
- Quality Document -- Project 06 Capstone
- data
- sample-documents
- shared
- data
- sample-documents
- data
- Skills
- Skills(技能集)
- Skills(技能集)
- harness-creator
- Harness Creator
- references
- templates
开始学习 → 拆解 Claude Code 的 harness 设计