← 返回 Skills 市场
Trinity Harness
作者
Christianye
· GitHub ↗
· v1.0.0
· MIT-0
93
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install trinity-harness
功能描述
Production-grade Agent Harness combining execution discipline (Superpower), knowledge compounding (CE), and product thinking (Gstack) into a single adaptive...
安全使用建议
This is an instruction-only 'engineering harness' that appears internally consistent and does not request credentials or downloads. Before installing, confirm the agent runtime's file and command-execution privileges are appropriately sandboxed: the harness instructs the agent to write checkpoints and run verification commands (ls, wc, grep, git blame, run tests). If the runtime can access sensitive files or external networks, restrict its filesystem scope and monitor outputs to prevent accidental exposure. If you need higher assurance, run the skill in a disposable/sandbox agent instance and review any generated checkpoints or logs.
功能分析
Type: OpenClaw Skill
Name: trinity-harness
Version: 1.0.0
The 'trinity-harness' skill bundle is a set of high-level engineering guidelines and workflow instructions designed to improve the reliability and discipline of an AI agent. It contains no executable code, only markdown documentation (SKILL.md and reference files) that promotes rigorous testing, verification, and checkpointing while explicitly warning against common pitfalls like unauthorized refactoring or hiding failures.
能力标签
能力评估
Purpose & Capability
Name/description (agent harness for engineering workflows) matches the SKILL.md content. There are no unexpected binaries, credentials, or unrelated requirements.
Instruction Scope
The instructions assume the agent can read/write files, run basic verification commands (ls, wc, grep, git blame, run tests) and persist checkpoints to the filesystem. This is appropriate for a harness, but it does require filesystem and command execution privileges in the agent runtime — verify those are scoped/sandboxed in your environment.
Install Mechanism
No install spec and no code files beyond documentation — lowest-risk delivery model (instruction-only). Nothing is downloaded or installed.
Credentials
The skill requests no environment variables, credentials, or config paths. Declared requirements align with the described functionality.
Persistence & Privilege
always:false and user-invocable:true (defaults). The skill does not request elevated persistence or modify other skills' configs.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install trinity-harness - 安装完成后，直接呼叫该 Skill 的名称或使用
/trinity-harness触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release: Three-layer adaptive harness — Challenge + Execute + Compound. Anti-rationalization guards, concurrent scheduling, MAST failure taxonomy, checkpoint patterns. Born from 450K words of AI textbook + 9 real production incidents.
元数据
常见问题
Trinity Harness 是什么？
Production-grade Agent Harness combining execution discipline (Superpower), knowledge compounding (CE), and product thinking (Gstack) into a single adaptive... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 93 次。
如何安装 Trinity Harness？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install trinity-harness」即可一键安装，无需额外配置。
Trinity Harness 是免费的吗？
是的，Trinity Harness 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Trinity Harness 支持哪些平台？
Trinity Harness 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Trinity Harness？
由 Christianye（@christianye）开发并维护，当前版本 v1.0.0。
推荐 Skills