The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from external, untrusted URLs to capture screenshots, creating a surface for indirect prompt injection if the resulting image or data is interpreted by the agent.
Ingestion points: Target URLs provided to the screenshot capture tools in SKILL.md.
Boundary markers: No specific delimiters or warnings for the agent to ignore instructions embedded in the target web content are provided in the instructions.
Capability inventory: Captures screenshots using strale_screenshot_url and context_dev_capture_screenshot via the sandbase_call_tool interface.
Sanitization: No explicit sanitization or filtering of the captured content is described.
[SAFE]: The skill uses a structured gateway (sandbase_describe_tool and sandbase_call_tool) to interact with specific screenshot capabilities, which provides a layer of controlled execution.
[SAFE]: The tools identified (strale_screenshot_url, context_dev_capture_screenshot) are part of the SandBase ecosystem associated with the skill's author and represent intended functionality for visual monitoring and archival.

screenshot-capture