doc-review
Skill0 starsChanged 2 months agoMITView on GitHub
单据/材料/报销的智能审核与置信度分流——规则引擎做交叉校验，确定单自动直通，低置信与问题单进人工队列，误放必须为零。当用户要"自动审核材料/发票/办件/包装"时使用。
15 other projects use a version of this.
Adopted by
Most recently taken up by
- momentmaker/kaijutsutheir own version, first seen Nov 1, 2026
- mbaneshi/claude-code-masterytheir own version, first seen Oct 20, 2026
- rterakedis/leanwheel-skillstheir own version, first seen Oct 17, 2026
- maaferna/job-search-agent-systemtheir own version, first seen Oct 15, 2026
- by-carlos/daikenjatheir own version, first seen Oct 13, 2026
- sd0xdev/sd0x-dev-flowtheir own version, first seen Oct 8, 2026
Other versions15 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 15
- stacklok/toolhive★ 2.3ksame name
- ffroliva/gflow-cli★ 367same name
- sd0xdev/sd0x-dev-flow★ 192same name
- sd0xdev/sd0x-harness★ 192same name
- YougLin-dev/Aha-Loop★ 181same name
- tlzmw001/aitest-kit★ 13same name
- ichangting/taotao-skills★ 8same name
- momentmaker/kaijutsu★ 3same name
- clearsmog/claude-skills★ 2same name
- rterakedis/leanwheel-skills★ 2same name
5 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository10 other files it gives its agents.
- llms.txtllms.txt
- content-pipelineSkill
- cost-quoteSkill
- fde-discoverySkill
- kb-buildSkill
- recon-chainSkill
- record-linkSkill
- semantic-layerSkill
- state-reminderSkill
- timeline-checkSkill
The file
Files
Doc Review · 单据审核与置信分流
对应案例：NO.2（车管所办件）、NO.9（央企财务报销）、NO.19（包装审核）。
参考实现：cases/02-doc-review/（零漏放零误拦，产能 15×人工）、
cases/09-finance-audit/（检出 100%、误放 0、直通 78%）。
何时使用
- 人工逐张审核材料/发票/办件/包装稿，量大、规则明确、错放后果严重
- 要把"人工全审"改成"AI 直通 + 人工只看风险"
流程
- 必填与齐全性（V1 基线）：必需材料清单 + 必填字段非空——这是最小可上线版本。
- 交叉校验（V2）：证照校验位算法、跨材料字段一致性（姓名/金额，只比对 语义上应该一致的材料——保单金额≠车价）、有效期。
- 双角色复核：执行角色与监督角色独立按规则集复核，结论交叉 （案例 9 的"一个执行判断、一个监督、一个发现风险"）。
- 置信度分流：抽取置信度 < 阈值（默认 0.9）→ review，人工只看问题材料； 硬错误 → reject 并附依据；无任何问题 → auto_pass。
- 红线纪律：误放（有问题却直通）= 0 是硬约束；为守住红线牺牲直通率是 正确取舍。
- 尺度问题上交：审核发现的历史"放宽"行为，把数据与影响摆给甲方决策 （案例 2 的核心转折），技能不替业务定尺度。
验收
- 植入缺陷的测试集上：误放 = 0，检出率 ≥ 95%
- 干净单据误拦率有上限（案例 2 口径：0）
- 每条 reject/review 附可解释依据
边界
- 图像抽取（OCR/多模态）是上游，本技能假设字段与置信度已产出
- 涉及业务利益的审核尺度决策不自动化
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.