fuse-browser-usage
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill mentions that its Chromium binary is obtained via an external script (setup.sh). As this script is not part of the skill package, its source and the integrity of the downloaded binary are unverifiable.
- [COMMAND_EXECUTION]: The documentation references the execution of a shell script (setup.sh) for environment setup.
- [PROMPT_INJECTION]: The skill is designed to ingest and process data from external websites, creating a surface for indirect prompt injection attacks. * Ingestion points: Untrusted data is retrieved through tools such as browser_fetch, browser_crawl, and browser_serp_batch (SKILL.md, references/research-docs.md). * Boundary markers: The instructions do not define delimiters or warnings to ignore instructions embedded in fetched content. * Capability inventory: The skill grants the agent interactive capabilities including browser_act (natural language browser control), browser_navigate, browser_console, browser_network, and browser_cookies (references/webapp-testing.md). * Sanitization: There are no requirements for sanitizing or validating external content before the agent processes it.
- [DATA_EXFILTRATION]: The agent's ability to access sensitive session data (e.g., via browser_cookies) while concurrently processing untrusted web content creates a risk of data exfiltration if the agent follows malicious instructions hidden on a webpage.
Audit Metadata