adminlove520/harness-creator
harness-creator/SKILL.md
创建生产级 Harness 脚手架的技能。 用于快速为一个新项目或工作区创建 AGENTS.md、feature_list.json、init.sh、claude-progress.md 等核心文件。 基于 WalkingLabs Learn Harness Engineering 课程。
$ install --global
skillsauthnpx skillsauth add adminlove520/xiaoxi-skills harness-creatorInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Apr 30, 2026, 3:17 AM8.8s1 file scanned
SKILL.md
- name:
- harness-creator
- description:
- >
Harness Creator
功能
快速为一个项目创建 Harness Engineering 脚手架文件。
使用场景
- 新项目开工
- 当前项目缺少 AGENTS.md 或其他 harness 文件
- 需要标准化 agent 工作流程
创建的文件
AGENTS.md
根指令文件。定义：
- 开工流程（开工前读什么）
- 工作规则（做什么、禁止做什么）
- 完成定义（什么叫"做完"）
- 验证标准
feature_list.json
机器可读的功能清单。每个功能包含：
- id、priority、area、title
- user_visible_behavior
- status (not_started/in_progress/blocked/passing)
- verification 步骤
- evidence (验证通过后记录)
init.sh
启动脚本：
- INSTALL_CMD - 依赖安装
- VERIFY_CMD - 基础验证
- START_CMD - 开发服务器启动
claude-progress.md
进度日志。记录：
- 当前已验证状态
- 当前最高优先级
- 会话记录
使用方法
对项目目录运行：
# 在项目根目录执行
cat > AGENTS.md << 'EOF'
# 项目名
## 开工流程
1. 读取 feature_list.json 了解当前状态
2. 读取 claude-progress.md 了解进度
3. 检查未完成功能的验证状态
## 完成定义
- [ ] 测试通过
- [ ] 代码通过 lint
- [ ] 文档已更新
EOF
cat > feature_list.json << 'EOF'
{
"features": []
}
EOF
五层防御检查清单
创建 harness 时，确保：
- 任务规范层 - 完成定义是否明确可验证？
- 上下文供给层 - feature_list 是否包含所有关键信息？
- 执行环境层 - init.sh 是否覆盖依赖安装？
- 验证反馈层 - 每项功能是否有验证步骤？
- 状态管理层 - claude-progress 是否记录跨会话状态？
参考
- https://walkinglabs.github.io/learn-harness-engineering/zh/resources/templates/
- OpenAI: Harness engineering
- Anthropic: Effective harnesses for long-running agents
Related Skills
adminlove520/memento-flashcards
data-ai
VerifiedTrustedCommunity
Spaced-repetition flashcard system. Create cards from facts or text, chat with flashcards using free-text answers graded by the agent, generate quizzes from YouTube transcripts, review due cards with adaptive scheduling, and export/import decks as CSV.
5SKILL.mdUpdated Apr 30, 2026
adminlove520/canvas
development
VerifiedTrustedCommunity
Canvas LMS integration — fetch enrolled courses and assignments using API token authentication.
5SKILL.mdUpdated Apr 30, 2026
adminlove520/distributed-llm-pretraining-torchtitan
development
VerifiedTrustedCommunity
Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing.
5SKILL.mdUpdated Apr 30, 2026
adminlove520/tensorrt-llm
devops
VerifiedTrustedCommunity
Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency. Use for production deployment on NVIDIA GPUs (A100/H100), when you need 10-100x faster inference than PyTorch, or for serving models with quantization (FP8/INT4), in-flight batching, and multi-GPU scaling.
5SKILL.mdUpdated Apr 30, 2026