The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation recommends installing the snapshot-query utility from the Python Package Index (PyPI), which is a well-known package registry.
[REMOTE_CODE_EXECUTION]: The workflow involves using uvx to execute the snapshot-query tool, which downloads and runs external code at runtime.
[COMMAND_EXECUTION]: The skill utilizes shell commands, including grep for searching snapshot files and uvx for tool execution, as part of its core functionality.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and acts upon untrusted data from web pages and OCR results. Ingestion points: Web content retrieved via browser_navigate and browser_snapshot, and OCR results from mcp_fast-paddleocr-mcp_ocr_image. Boundary markers: No explicit delimiters or instructions are used to isolate untrusted web content from the agent's primary instructions. Capability inventory: The skill provides extensive browser control, including browser_click, browser_type (for text input), and browser_navigate. Sanitization: There is no evidence of sanitization or filtering of the ingested web content before it is processed by the agent.

cursor-ide-browser-skills