The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install the open-computer-use package globally using npm install -g.
[EXTERNAL_DOWNLOADS]: The skill provides instructions to install the cli-anything-hub package via pip install.
[REMOTE_CODE_EXECUTION]: The Playwright configuration uses npx -y @playwright/mcp, which fetches and executes code from Microsoft's Playwright repository on the npm registry.
[COMMAND_EXECUTION]: The skill requires the user to execute shell commands for software installation and to manually modify the local opencode.json configuration file.
[PROMPT_INJECTION]: The skill creates an Indirect Prompt Injection surface by installing tools that ingest and act upon untrusted external data.
Ingestion points: The Playwright and open-computer-use tools retrieve content from arbitrary URLs and desktop environments.
Boundary markers: No boundary markers or delimiters are specified to help the agent distinguish between retrieved content and system instructions.
Capability inventory: The tools provide the ability to automate browser actions, take screenshots, and interact with the user's desktop.
Sanitization: There is no evidence of sanitization or filtering of the external data before it is processed by the agent.

opencode-browser