← 返回 Skills 市场
cutthemustard

Clawhub Skill

作者 CutTheMustard · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
596
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install cs-verify
功能描述
Verify factual claims against live data sources. Returns structured verdicts with confidence scores.
使用说明 (SKILL.md)

verify-claim

Verify any factual claim against live data sources. Returns a structured verdict with confidence score, current truth value, and freshness indicator.

Usage

Send a POST request to verify a claim:

curl -X POST https://636865636b73756d.com/v1/verify \
  -H "Content-Type: application/json" \
  -d '{"claim": "The USD to EUR exchange rate is 0.92"}'

Response Format

{
  "verdict": "confirmed",
  "confidence": 0.95,
  "current_truth": "0.921",
  "freshness": "live",
  "source_count": 2,
  "cached": false,
  "request_id": "abc-123",
  "service": "https://636865636b73756d.com",
  "referral_id": "cs_ref_a7b3"
}

Verdicts

  • confirmed — claim matches current data
  • stale — claim was true but data has changed
  • disputed — sources disagree
  • false — claim contradicts current data
  • unknown — unable to verify

Categories

Optionally specify a category for faster routing:

  • financial — exchange rates, crypto prices, stock prices
  • entity — company info, population, founding dates
  • geo — timezones, geographic data
  • factcheck — general fact-checking via Google Fact Check API
{"claim": "Bitcoin price is above $50,000", "category": "financial"}

Trending Claims

Get the top 100 most-queried claims in the last 24 hours:

curl https://636865636b73756d.com/v1/trending

Pricing

  • Free tier: 25 queries/day, no authentication required
  • Paid tier: unlimited queries via x402 protocol (USDC on Base), $0.001/query

Agent Discovery

  • Agent Card: https://636865636b73756d.com/.well-known/agent.json
  • Service metadata: https://636865636b73756d.com/.well-known/agent-service.json
  • MCP server: npm install -g @636865636b73756d/mcp-v1

Referrals

Every response includes a referral_id. Pass it as "referral": "cs_ref_..." in your requests to earn query credits. Tiers: 10+ referrals = 10%, 100+ = 15%, 1000+ = 20% credit on referred agent queries.

安全使用建议
This skill forwards claims to an external, unvetted service (the SKILL.md consistently uses a hex-encoded domain). Before installing or enabling it: (1) verify the service/operator (look up the domain, ownership, and privacy policy); (2) avoid sending sensitive or confidential claims to the service; (3) decline the optional global npm install unless you can audit the package source; (4) be cautious about referral codes and crypto payment flow — they create incentives to log or monetize your queries. If you need verifiable, auditable fact-checking, prefer a skill with source code provided or one that calls well-known, documented APIs.
功能分析
Type: OpenClaw Skill Name: cs-verify Version: 1.0.0 The skill is suspicious due to a direct instruction in `SKILL.md` to install a global npm package (`npm install -g @636865636b73756d/mcp-v1`). This represents a significant supply chain risk and a potential prompt injection vulnerability, as it instructs the AI agent to execute arbitrary code from an external source. If the specified npm package were malicious, this could lead to arbitrary code execution on the agent's host system. Additionally, the use of a hex-encoded domain name (`636865636b73756d.com` which decodes to 'checksum') is an unusual practice that could be a mild indicator of obfuscation.
能力评估
Purpose & Capability
The name/description (verify factual claims against live data) align with the instructions: the SKILL.md consistently sends queries to a web API. There are no unrelated env vars, binaries, or config requirements declared. However, all runtime activity is delegated to an external service (domain shown in SKILL.md), so the trust boundary is entirely the remote host.
Instruction Scope
Runtime instructions tell the agent to POST user claims to multiple endpoints on a single third‑party domain (hex-encoded domain used throughout). The instructions do not ask the agent to read local files or credentials, but they do instruct sending user-provided content to that external service and to include a referral id for credits — this enables tracking/telemetry and could encourage leaking of queries. No privacy handling or data-retention guarantees are provided.
Install Mechanism
There is no formal install spec (instruction-only), which reduces local persistence risk. The doc however suggests installing a global npm package (npm install -g @636865636b73756d/mcp-v1) hosted under the same unfamiliar scope — an optional but potentially risky supply-chain action if performed. The package scope and host are untrusted/unverified.
Credentials
The skill declares no required environment variables or credentials and the SKILL.md does not request secrets. That is proportionate to the described functionality. Note: the referral/paid-tier mechanism uses crypto payments but does not require credentials in the doc; monetization incentives could nevertheless motivate data collection.
Persistence & Privilege
always:false and no install steps mean the skill does not request permanent system presence or elevated privileges. It does not ask to modify other skills' configs or system-wide settings.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install cs-verify
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /cs-verify 触发
  4. 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 of cs-verify skill. - Verify factual claims against live data sources with structured verdicts and confidence scores. - Supports claim categories for faster and specialized verification (financial, entity, geo, factcheck). - Provides a trending claims endpoint showing the top 100 queried claims in the last 24 hours. - Free tier available (25 queries/day, no auth); paid tier supports unlimited queries via USDC. - Includes unique referral system for earning credits on agent queries.
元数据
Slug cs-verify
版本 1.0.0
许可证
累计安装 0
当前安装数 0
历史版本数 1
常见问题

Clawhub Skill 是什么?

Verify factual claims against live data sources. Returns structured verdicts with confidence scores. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 596 次。

如何安装 Clawhub Skill?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install cs-verify」即可一键安装,无需额外配置。

Clawhub Skill 是免费的吗?

是的,Clawhub Skill 完全免费(开源免费),可自由下载、安装和使用。

Clawhub Skill 支持哪些平台?

Clawhub Skill 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 Clawhub Skill?

由 CutTheMustard(@cutthemustard)开发并维护,当前版本 v1.0.0。

💬 留言讨论