camoufox-2026
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the download of a pre-built, patched Firefox binary from the official Camoufox project repository on GitHub.
- Evidence: The command
python -m camoufox fetchis used to retrieve a ~300MB binary fromhttps://github.com/daijro/camoufox/releases. - [COMMAND_EXECUTION]: The skill requires executing local setup commands to install dependencies and browser binaries.
- Evidence: Installation steps include
uv add camoufox[geoip]andpython -m camoufox fetchto prepare the execution environment. - [INDIRECT_PROMPT_INJECTION]: As a browser automation skill, it possesses an inherent attack surface for indirect prompt injection by navigating to and reading content from external, untrusted websites.
- Ingestion points: The skill uses
page.goto(url)to load external web content into the agent's context. - Capability inventory: The skill has full browser automation capabilities, including network navigation, cookie management, and screenshot saving (
page.screenshot). - Boundary markers: The provided examples do not explicitly define boundary markers or delimiters for the content retrieved from websites.
- Sanitization: No explicit sanitization or filtering of the HTML/text content retrieved from the browser is demonstrated in the provided code snippets.
Audit Metadata