The Agent Skills Directory

[PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection, as it ingests untrusted content from the web and uses it to perform local file system modifications.
Ingestion points: The agent navigates to remote URLs and captures DOM structure/snapshots using browser_navigate and browser_snapshot tools as described in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions potentially hidden within the target website's HTML or metadata.
Capability inventory: The skill possesses significant capabilities, including File Read/Write for applying code fixes and Web Page Navigation for traversing external sites.
Sanitization: No sanitization or validation logic is specified for the data retrieved from external URLs before it influences the agent's code-writing actions.
[EXTERNAL_DOWNLOADS]: The skill's implementation guidance suggests the use of the Playwright MCP server from Microsoft.
It recommends executing @playwright/mcp via npx to enable browser automation capabilities like vision and DOM snapshots.
[COMMAND_EXECUTION]: The workflow relies on executing browser automation commands to interact with and inspect live web applications.
The skill utilizes tools such as browser_navigate, browser_resize, and browser_take_screenshot to perform visual and responsive design testing.

web-design-reviewer