dev-browser
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches additional skill components and reference documentation from the author's GitHub repository (
Melvynx/ai-maxxing-skills) usingcurlas required by the task. - [COMMAND_EXECUTION]: Utilizes Bash commands to manage the browser environment (
chrome-ai) and execute automation tasks via thedev-browserCLI. - [DYNAMIC_EXECUTION]: Programs the agent to generate and execute JavaScript code within a sandboxed QuickJS environment to interact with web pages.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from websites, which presents a surface for indirect prompt injection attacks.
- Ingestion points: Website content and snapshots retrieved via
page.snapshotForAI()inSKILL.md. - Boundary markers: None identified; web content is processed without explicit separation from instructions.
- Capability inventory: Includes the ability to navigate URLs, click page elements, fill forms, and write files to the
~/.dev-browser/tmp/directory. - Sanitization: No sanitization or validation of external web content is specified before processing.
Audit Metadata