opencli-browser
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the Bash(opencli:*) tool to manage browser sessions and drive interactions. This involves executing subcommands for navigation, element selection, and state retrieval.
- [DYNAMIC_EXECUTION]: The browser eval command facilitates the execution of JavaScript within the browser's context. The instructions recommend using this for read-only data extraction and specifically warn against using it for page modifications, which should be handled by dedicated interaction commands.
- [DATA_EXPOSURE]: The skill describes tools for accessing sensitive browser data, such as capturing network traffic (browser network), taking screenshots (browser screenshot), and uploading local files to the browser (browser upload). These are standard features for browser automation tools.
- [INDIRECT_PROMPT_INJECTION]: Because the skill retrieves and processes data from external websites, it possesses an attack surface for indirect prompt injection. 1. Ingestion points: Page content is ingested through the browser state, get html, and network commands. 2. Boundary markers: There are no instructions for using delimiters to isolate untrusted web content from agent prompts. 3. Capability inventory: The opencli tool can perform a wide range of browser actions, including clicking, typing, and executing scripts. 4. Sanitization: No mechanisms are described for sanitizing or escaping the data retrieved from the web before it is analyzed by the agent.
Audit Metadata