Quality Pipeline
v1.1.0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。
⭐ 1· 94·1 current·1 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & 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.
Assessment
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.Like a lobster shell, security has layers — review code before you run it.
latestvk9752x5cbh98ss9bdfx4as6j0183w499
License
MIT-0
Free to use, modify, and redistribute. No attribution required.