← 返回 Skills 市场
141
总下载
1
收藏
1
当前安装
2
版本数
在 OpenClaw 中安装
/install quality-pipeline
功能描述
产出物多层级交叉质检流水线。任务完成后自动触发，干活的agent不能审自己，换agent交叉验证。适用场景：论文终稿、公文成品、报告文档等所有正式产出物。触发词：质检、验收、四层验证、交叉检查、质检流水线。
安全使用建议
This skill appears coherent and focused on document QA, but before installing consider: 1) The pipeline will read full document contents you pass in—do not feed confidential material unless you trust the agent environment. 2) DOI and Google Scholar checks imply the agent may perform web queries or use retrieval tools; confirm what network access or external tools your agent will use. 3) 'ThunderVault' and model names in the docs are operational references—verify whether they map to services in your environment and whether credentials are required. 4) If you want tighter safety, restrict the agent's network access or disable autonomous invocation until you confirm the toolchain used for external lookups. Overall, functionality, scope, and requested privileges match the stated purpose.
能力评估
Purpose & Capability
Name/description (multi-layer QA for papers/reports) align with the SKILL.md: the document describes L1–L5 checks, cross-agent review, DOI/data verification, and return-for-fix rules. The skill requests no unrelated binaries, env vars, or config paths. Model/agent names in the docs appear as labels/recommendations rather than declared dependencies.
Instruction Scope
Runtime instructions ask agents to open and inspect files at a provided {path}, validate DOIs/titles via search/Google Scholar cross-checks, and use multiple agents/models. These behaviors are appropriate for the stated QA purpose, but they imply that the agent will read full document contents and may perform web queries or tool-based searches (not declared in metadata). Users should be aware the pipeline requires access to the document contents and may invoke external lookups during verification.
Install Mechanism
Instruction-only skill with no install spec and no code files. Nothing is written to disk by the skill itself and no external packages are fetched by an installer.
Credentials
The skill declares no required environment variables, credentials, or config paths. Suggested verification steps reference external services (DOI check, Google Scholar, 'ThunderVault' path), but the skill does not request credentials for them; network/tool usage would come from the agent environment, not the skill manifest.
Persistence & Privilege
always:false and no actions that modify other skills or system settings. disable-model-invocation is false (normal). The skill does instruct spawning multiple agents/models at runtime, which is coherent with the QA workflow but increases runtime activity surface.
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 quality-pipeline - 安装完成后，直接呼叫该 Skill 的名称或使用
/quality-pipeline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
v1.1.0
Added writing-standards.md (de-AI rules, format specs, scoring rubric) + statistical-tables.md (8 standard tables + Kano-IPA templates)
v1.0.0
Initial release: 5-layer cross-review pipeline for deliverables
元数据
常见问题
Quality Pipeline 是什么？
产出物多层级交叉质检流水线。任务完成后自动触发，干活的agent不能审自己，换agent交叉验证。适用场景：论文终稿、公文成品、报告文档等所有正式产出物。触发词：质检、验收、四层验证、交叉检查、质检流水线。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 141 次。
如何安装 Quality Pipeline？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install quality-pipeline」即可一键安装，无需额外配置。
Quality Pipeline 是免费的吗？
是的，Quality Pipeline 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Quality Pipeline 支持哪些平台？
Quality Pipeline 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 Quality Pipeline？
由 luaqnyin（@luaqnyin）开发并维护，当前版本 v1.1.0。
推荐 Skills