Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed third-party web search skill; it sends queries to an external provider, but that behavior matches its stated purpose and includes privacy warnings and limits.
Install only if you are comfortable sending search queries to clb.ciglobal.cn with an account-linked API key. Do not use it for passwords, tokens, personal identifiers, confidential business data, or internal-only content; use redacted/generalized queries and verify the provider's privacy and retention practices first.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
import os
API_URL = "https://clb.ciglobal.cn/web_search"
# Get API key from environment variable (recommended)
API_KEY = os.environ.get("GLOBAL_SEARCH_API_KEY", "your_api_key_here")
if API_KEY == "your_api_key_here":
The cURL example demonstrates direct transmission of search terms and an API key to an external third-party service. Although documented openly, this is still a true external transmission issue because it normalizes sending potentially sensitive user content off-platform and could expose account-linked search history to the provider.
curl -X POST "https://clb.ciglobal.cn/web_search" \
The JavaScript example sends user search queries and an API key to a third-party endpoint over the network. In the context of an agent skill, this creates a real privacy and data-governance risk because user inputs may contain sensitive data, and the provider may log queries and associate them with the caller's account.
fetch('https://clb.ciglobal.cn/web_search', {
method: 'POST',
headers: {
'X-API-Key': 'your_api_key_here', // Get your API key at https://clb.ciglobal.cn/apiKey/login
The skill description markets broad, default multi-source web search for general research tasks, which can cause the agent to invoke an external third-party service for many ordinary information requests without sufficient narrowing. In this context, overly broad activation increases the chance of unnecessary external transmission of user prompts, including potentially sensitive or internal data, despite the later privacy warnings.
The trigger conditions are ambiguous enough to match common requests like 'gather information' or 'retrieve references,' which may cause the skill to be selected by default rather than using safer local reasoning or narrower tools. Because this skill transmits queries to an external provider, overbroad routing materially raises privacy and data-governance risk.
The skill explicitly instructs sending user search terms to an external endpoint controlled by a third party. Even though the file includes warnings and consent guidance, the core behavior still creates a real data-exfiltration surface if users or agents submit sensitive, confidential, or regulated information in queries.
Body (form): keyword=机器学习&mode=warehouse&page=1
### Example 4: cURL
```bash
curl -X POST "https://clb.ciglobal.cn/web_search" \
-H "X-API-Key: <your_api_key>" \
The skill contains a mandatory instruction to use this script whenever a user requests a comprehensive search, without presenting any language, locale, or provider choice. In context, this increases the chance that user queries are automatically sent to a third-party Chinese service, which can create privacy, consent, and jurisdictional risks, especially for users who did not intend cross-border or provider-specific disclosure.
The document mixes English with Chinese labels and instructions, including operational guidance such as the comprehensive-search note in Chinese. This imposes locale-specific content without stating that the skill is region-specific or giving the user an option to choose language/locale.