The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the zerotoken Python package and Playwright browser binaries (playwright install chromium). These are legitimate dependencies for the skill's functionality and originate from the vendor (AMOS144) or well-known services.
[COMMAND_EXECUTION]: The skill executes browser automation commands via MCP tools (e.g., browser_click, browser_input) and manages background processes for the MCP server (zerotoken-mcp-http). It also utilizes a custom script engine to replay recorded sequences of actions.
[PROMPT_INJECTION]: The skill processes untrusted external data from websites using tools like browser_get_text and browser_extract_data. This creates a surface for indirect prompt injection, as malicious instructions embedded in a website could attempt to influence the agent's logic. No explicit sanitization or boundary markers are defined for this ingested content.

zerotoken-openclaw