Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed AdMapix API helper that uses an API key and local caching for advertising-data lookups, with some documentation cleanup needed but no artifact-backed malicious behavior.
Install only if you intend to send advertising-analysis queries to AdMapix and are comfortable using an AdMapix API key from the environment. Keep the key out of chat, review any requested exports or cache writes, and choose explicit local paths for saved data.
The usage section explicitly states '原始结构化 JSON 透传，不重命名、不丢弃、不总结、不排序、不评论', and the capability boundary later says '不做分析与推荐'. However, the examples immediately below present transformed output objects and narrative analyses such as trend judgments and risk prompts, which contradict the stated no-summary/no-analysis behavior.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# Step 1: 搜索竞品产品
curl -s -X POST "https://api.admapix.com/api/data/product-search" \
-H "X-API-Key: ${ADMAPIX_API_KEY}" -H "Content-Type: application/json" \
-d '{"keyword":"竞品名称","page":1,"page_size":10}'
# ...
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# Step 1: 搜索竞品产品
curl -s -X POST "https://api.admapix.com/api/data/product-search" \
-H "X-API-Key: ${ADMAPIX_API_KEY}" -H "Content-Type: application/json" \
-d '{"keyword":"竞品名称","page":1,"page_size":10}'
# ...
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# Step 1: 搜索竞品产品
curl -s -X POST "https://api.admapix.com/api/data/product-search" \
-H "X-API-Key: ${ADMAPIX_API_KEY}" -H "Content-Type: application/json" \
-d '{"keyword":"竞品名称","page":1,"page_size":10}'
# ...
The documentation instructs writing paginated results to a local file but does not provide a clear warning or consent requirement for modifying local data. In an agent setting, even benign export guidance can lead to silent persistence of potentially sensitive or bulky data on disk without user awareness.
The capability boundary says '单次请求单端点...不内置批量端点聚合', implying a thin client without built-in orchestration. This conflicts with earlier sections that present endpoint dependency orchestration as a core capability and describe automatic parallel/serial execution across multiple endpoints in one round.
Advertising file write operations without a strong filesystem safety warning can normalize local modifications beyond the user’s expectations. In agent environments, this raises the risk of unintended overwrites, persistence of third-party data, or writing into sensitive directories.
The skill is positioned as an AdMapix data-access/orchestration tool, but the documented feature set expands into generic file handling and command execution. Broadening scope in a skill that has exec/read/glob/grep available increases the chance an agent will perform unnecessary local actions, including writing files or running shell commands outside the expected data-retrieval workflow.