screenshot-maker
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to interact with and extract data from external, untrusted web pages, which could contain malicious instructions intended to hijack the agent's behavior.
- Ingestion points: The skill navigates to user-provided URLs and inspects page structures using
agent-browseras documented inSKILL.mdand implemented inscripts/prepare.py. - Boundary markers: The instructions do not specify any markers or guidelines for the agent to distinguish between its core instructions and potentially malicious content found on the web pages it visits.
- Capability inventory: Across its scripts, the skill can execute system commands (
magick,agent-browser) and perform file operations within the/tmp/directory. - Sanitization: The implementation in
scripts/prepare.pycorrectly employsjson.dumps()when interpolating selectors and text into JavaScript payloads, providing protection against script injection within the browser environment. - [COMMAND_EXECUTION]: The skill relies on executing external CLI tools to perform image processing and browser automation tasks.
- Evidence: Functions in
scripts/frame.py(e.g.,magick,identify) andscripts/prepare.py(e.g.,run_ab,eval_js,is_blank_image) usesubprocess.run()to call external executables. - Context: These calls use list-based argument passing and avoid the use of a shell (
shell=True), which significantly reduces the risk of command injection from maliciously crafted inputs.
Audit Metadata