The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions direct the installation of the smooth-py Python package. This package is the official SDK for the Smooth.sh cloud service utilized by the skill.
[REMOTE_CODE_EXECUTION]: The evaluate-js command enables the agent to execute arbitrary JavaScript code within the browser session. While a standard feature for automation, this provides a mechanism for dynamic code execution in the browser environment.
[DATA_EXFILTRATION]: The skill provides functionality to upload local files to the Smooth.sh cloud environment via the upload-file command. It also allows for the retrieval of data through the downloads command. Users should be aware that data uploaded through these commands is processed on the vendor's infrastructure.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and process content from arbitrary third-party websites.
Ingestion points: Web content is retrieved and processed by the agent during smooth run and smooth extract operations.
Boundary markers: No explicit boundary markers or instructions are provided to help the agent distinguish between its system-level instructions and instructions that may be embedded in external web content.
Capability inventory: The agent has capabilities to execute shell commands, run JavaScript in the browser, and perform file upload/download operations.
Sanitization: The skill documentation does not mention any sanitization or validation of content fetched from external URLs before it is processed by the agent.

smooth-browser