Contract Check

v1.0.1

通用商务合同审核与风险检测技能。安装后必须先引导用户完成企业商业底线配置,之后才能审核合同。支持PDF/Word/文本格式合同审核,输出结构化审核报告,包含否决项、警告项和改进建议。当用户要求审核合同、检查合同风险、审查合同条款、合同自审、合同红线检查时激活。

0· 195·0 current·0 all-time
byChartGen AI@chartgen-ai·duplicate of @chartgen-ai/contracts-review

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for chartgen-ai/contract-check.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Contract Check" (chartgen-ai/contract-check) from ClawHub.
Skill page: https://clawhub.ai/chartgen-ai/contract-check
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install contract-check

ClawHub CLI

Package manager switcher

npx clawhub@latest install contract-check
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
Name/description (contract review, redline checks) match the actual behavior: reading reference templates, asking configuration questions, scanning contract text against configured redlines and keywords. No unrelated credentials, binaries, or external services are requested.
Instruction Scope
SKILL.md stays within the contract-review domain (initialization, per-review checks, redline rules). It explicitly instructs reading/writing a single config file under ~/.openclaw/workspace/.contract-review-config.json and reading bundled reference files. One runtime command example uses exec (cat) to check file existence; the command is static and limited in scope. Recommendation: be aware the skill will persistently store business configuration and contract-related data locally.
Install Mechanism
Instruction-only skill with no install spec, no external downloads, and no code files to execute. This is the lowest-risk install posture and matches the content.
Credentials
The skill requests no environment variables, credentials, or external endpoints. It does write and read a local config file containing potentially sensitive company/business rules (company name, financial thresholds, IP policy). Requesting local file read/write access is proportional to purpose but users should recognize that sensitive business policy data will be stored on disk.
Persistence & Privilege
The skill persists its own configuration at ~/.openclaw/workspace/.contract-review-config.json and provides commands to view/modify/reset it. 'always' is false and the skill does not modify other skills or global settings. Persistent storage of configuration is expected for this use case but users should be aware of the local file presence and lifecycle.
Assessment
This skill appears internally consistent for reviewing contracts and enforcing per‑company redlines. Before installing, note: (1) it will create and read a persistent config file at ~/.openclaw/workspace/.contract-review-config.json — that file will contain company name, role (甲方/乙方) and business thresholds, so avoid storing secrets you wouldn't want on disk; (2) it runs a simple shell check (cat) to detect the config file — the command shown is static, but you should ensure the agent runtime environment is trusted and sandboxed; (3) the skill does not use network or request credentials, so data remains local unless your agent runtime explicitly forwards it elsewhere; (4) you can review and delete the config file via the skill's '重置配置' command if you want to remove persistent data. If you need stronger guarantees (encryption at rest, retention policy, or avoidance of writing persistent data), ask the publisher or avoid enabling persistence.

Like a lobster shell, security has layers — review code before you run it.

latestvk97fgx2ag37w7jms548qwanvvd8406rp
195downloads
0stars
2versions
Updated 3w ago
v1.0.1
MIT-0

通用合同审核技能

为企业提供可定制的商务合同系统化风险审查。

⛔ MANDATORY GATE — 初始化检查(不可跳过)

每次用户要求审核合同时,执行以下检查:

  1. exec 检查配置文件是否存在:
    cat ~/.openclaw/workspace/.contract-review-config.json 2>/dev/null || echo "NOT_FOUND"
    
  2. 如果返回 NOT_FOUND → 立即停止,进入【初始化流程】(见下方),禁止继续审核
  3. 如果配置存在 → 先确认用户的审核角色(甲方/乙方),再进入审核流程

🚫 绝对禁止:在配置文件不存在的情况下跳过初始化直接审核合同。 🚫 绝对禁止:用户未明确审核角色时自行假设。


初始化流程

当配置文件不存在时触发,必须按以下步骤执行:

1. 收集基础信息

一次性收集以下 3 项:

  • 审核角色:甲方(采购方)/ 乙方(服务/产品提供方)
  • 企业名称
  • 合同类型(可多选):产品销售 / 技术实施 / 定制开发 / 订阅(SaaS) / 混合型

💡 角色决定所有红线规则的适用方向,合同类型决定后续模块的自动跳过裁剪。

2. 选择配置模式

问用户:"选择配置模式:

  • 快速(1 个问题,其余用推荐默认值)
  • 详细(逐模块自定义,约 10-15 个问题)"

快速模式

只问 1 个核心问题:违约责任总额上限,不超过实际收款额的 ____%

  • 🔹 乙方建议 100%,甲方建议 100%-150%

回答后,读取 references/config-questions.md 中的「快速模式默认值」表,根据角色自动预填所有其余配置项,直接进入步骤 3 写入。

详细模式

读取 references/config-questions.md,按模块 1-7 依次向用户提问。

  • 跳过规则:根据合同类型选择,自动跳过不适用的问题
  • 甲/乙方适配:问题说明展示对应视角的建议值
  • 数据安全跳过:模块 5 先询问是否涉及数据处理,不涉及则跳过
  • 每次只问一个模块,等用户回答后再问下一个
  • 用户回答"按示例"时,使用问题中的示例值

3. 写入配置文件

所有配置确认后,将配置写入 ~/.openclaw/workspace/.contract-review-config.json

{
  "company": "企业名称",
  "review_role": "甲方 或 乙方",
  "contract_types": ["产品销售", "定制开发"],
  "config_mode": "quick 或 detailed",
  "version": "1.0.0",
  "created": "YYYY-MM-DD",
  "payment": {
    "first_payment_min_pct": 30,
    "first_payment_deadline_days": 10,
    "max_payment_cycle_days": 30
  },
  "liability": {
    "cap_pct_of_actual_receipt": 100,
    "compensation_scope": "仅限直接经济损失"
  },
  "ip": {
    "source_code_policy": "B",
    "source_code_ip_owner": "客户",
    "shared_ip_allowed": "仅限定制/共创功能部分"
  },
  "product_service": { ... },
  "confidentiality": { ... },
  "compliance": {
    "integrity_violation_penalty_pct": "5%-10%",
    "require_integrity_annex": "仅限政府及国企项目"
  },
  "service_rights": {
    "acceptance_penalty_daily_rate": 0.5,
    "acceptance_penalty_cap_pct": 3,
    "suspend_after_days": 30,
    "terminate_after_days": 60,
    "performance_bond_pct": 10
  }
}

配置完成后,询问用户"配置已完成!现在要审核合同吗?"

  • 如果用户确认 → 立即进入审核流程
  • 如果用户说稍后 → 回复"好的,随时发合同给我审核"

审核流程

配置存在时,按以下 5 步执行:

Step 1:预审确认

读取合同文本后,确认本次合同的关键信息(仅检查初始化中未覆盖的项目):

  1. 本次合同类型:____(对照配置中的 contract_types,若不在范围内则提示)
  2. 合同金额及与报价单差异
  3. 是否涉及第三方分包/渠道分佣(初始化未覆盖,需每次确认)
  4. 是否存在履约保证金要求(对照 service_rights.performance_bond_pct
  5. 是否涉及第三方知识产权授权(初始化未覆盖,需每次确认)

💡 定制开发、源代码交付、数据安全等项已在初始化时通过合同类型选择和模块配置完成,无需每次重复确认。仅在本次合同出现与配置不一致的情况时才需特别提示。

如有任一涉及,提示用户需提前评估风险。

Step 2:否决条件扫描(7条即时触发)

对合同文本逐一检查,命中任一条即输出【否决】

#检查项检查方式
1侵犯己方知识产权(共有IP超出配置范围)搜索"共同拥有""共同享有知识产权",对照 ip 配置
2无违约责任上限搜索"赔偿不封顶""无上限赔偿"
3付款周期超过配置最大天数提取付款条款日期,对照 payment.max_payment_cycle_days
4己方须提供所有源代码(超出配置范围)搜索"源代码""源码交付""全部源码",对照 ip 配置
5保密期限超过配置上限或永久搜索"永久保密""无期限保密",对照 confidentiality 配置
6无限连带赔偿责任(无上限)搜索"无限连带""连带赔偿"
7数据安全责任无限连带(无上限)搜索数据安全条款中的连带责任,对照 confidentiality 配置

Step 3:七大类红线逐项审查

按类别逐条对照审查。详细规则见 references/redlines.md,数值从配置文件读取:

  1. 知识产权 → 对照 ip
  2. 违约责任 → 对照 liability
  3. 验收&回款 → 对照 payment
  4. 产品&服务 → 对照 product_service
  5. 保密&数据安全 → 对照 confidentiality
  6. 合规&廉洁 → 对照 compliance
  7. 服务方权益 → 对照 service_rights

Step 4:必备条款完整性检查

确认合同是否包含以下必备条款(缺失为【警告】):

  • 违约责任上限(≤ 配置比例 × 实收费用)
  • 赔偿范围限定为配置范围(排除排除项)
  • 首款 ≥ 配置最低比例%(签署后配置期限内)
  • 付款周期 ≤ 配置最长天数
  • 验收环节 + 支付节点 + 验收标准附件
  • 验收逾期违约金(每天配置‰率,上限配置%)
  • 暂停服务权(逾期配置天数)
  • 送达条款(含电子送达)

订阅合同额外:订阅自动终止条款、扩容费用条款 买断合同额外:维保期限制、无"免费升级"承诺

Step 5:输出审核报告

📋 合同审核报告

基本信息:
- 审核角色:甲方/乙方
- 合同类型:xxx
- 合同金额:xxx
- 审核依据:企业自定义商业底线规则

🚫 否决项(N/N):
[命中条款+原文+修改建议]

⚠️ 警告项(N/N):
[风险点+对应条款+修改建议]

✅ 必备条款检查:
[逐项通过/缺失,显示配置值]

📝 改进建议:
[非红线但建议优化的条款]

审核结论:✅ 通过 / ⚠️ 有条件通过 / 🚫 否决

参考资料

文件用途何时读取
references/config-questions.md7 模块配置问答模板初始化时逐模块提问
references/redlines.md24 条红线详细规则Step 3 审查时读取
references/keywords.md关键词速查表Step 2 扫描时辅助定位

配置管理

操作触发方式说明
修改配置用户说"修改合同审核配置"可指定模块,如"修改合同审核配置:付款规则"
查看配置用户说"查看合同审核配置"显示当前完整配置
重置配置用户说"重置合同审核配置"删除配置文件,下次审核重新初始化
查看模块用户说"查看付款规则"等显示单个模块的当前配置值

Comments

Loading comments...