The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from external web pages, which could be used for indirect prompt injection attacks.
Ingestion points: Target website content is processed via tools like navigate_page and take_snapshot in SKILL.md.
Boundary markers: Absent. There are no instructions provided to the agent to treat website content as potentially untrusted or to ignore embedded instructions.
Capability inventory: The skill uses the chrome-devtools MCP server, allowing for browser navigation, trace analysis, and DOM snapshots.
Sanitization: Absent. The skill does not describe any sanitization or validation of the content retrieved from the web.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for a setup that involves fetching a package from a public registry at runtime.
Evidence: The configuration guide in SKILL.md suggests using the command npx -y chrome-devtools-mcp@latest.
Details: This command downloads and executes the latest version of an external Node.js package from the npm registry.

web-performance-audit