aliyun-modelstudio-crawl-and-skill
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses
npx -y @just-every/crawlto download and execute a utility from the NPM registry to perform web crawling. - [COMMAND_EXECUTION]: The workflow involves multiple shell command executions, including Node.js package execution (
npx), Python script execution for data processing, and filesystem operations (mkdir,py_compile). - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from an external documentation site (
help.aliyun.com) which is then used to generate new skill files and summaries. This represents an attack surface where maliciously crafted content on the source page could influence the generated skill instructions. - Ingestion points:
alicloud-model-studio-models.md(crawled content from Aliyun documentation). - Boundary markers: No specific boundary markers or 'ignore' instructions are used when interpolating crawled data into the generated summaries or skill reports.
- Capability inventory: The skill has permissions to write to the local filesystem, specifically targeting the
output/directory and theskills/ai/directory structure. - Sanitization: The Python scripts perform regex-based extraction and URL unquoting but do not sanitize the raw text for potentially malicious instructions or prompt-like patterns.
Audit Metadata