page-agent-web-automation
Pass
Audited by Gen Agent Trust Hub on May 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill utilizes the
page-agentlibrary, which is a legitimate open-source tool provided by Alibaba. It uses official distribution channels like NPM and the jsDelivr CDN. - [SAFE]: Documentation correctly encourages the use of environment variables (
process.env) for managing sensitive API keys for LLM providers, avoiding hardcoded credentials. - [SAFE]: No evidence of obfuscation, malicious network calls, or unauthorized persistence mechanisms was identified in the instructions or example code.
- [PROMPT_INJECTION]: The skill's primary function involves processing untrusted data from web page DOMs (ingestion point: SKILL.md), which presents a surface for indirect prompt injection. Capability inventory includes DOM manipulation and action execution via
agent.execute. While boundary markers are not explicitly defined in the provided snippets, the documentation includes a 'Best Practices' section recommending sanitization and validation of user input to mitigate risks.
Audit Metadata