The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted content from external URLs, which creates a surface for indirect prompt injection attacks where a website's content could attempt to influence the agent's behavior.
Ingestion points: The navigate_page(url: "...") tool call in Phase 1 (SKILL.md) and subsequent network/DOM analysis tools ingest external data into the agent's context.
Boundary markers: There are no explicit instructions or delimiters defined to separate untrusted website content from the agent's internal instructions.
Capability inventory: The skill uses a powerful set of tools including performance_start_trace, list_network_requests, get_network_request, and take_snapshot (SKILL.md, Quick Reference).
Sanitization: No specific sanitization or filtering of the ingested DOM or network data is mentioned before processing.
[EXTERNAL_DOWNLOADS]: The skill references and encourages the retrieval of information from trusted documentation sources and official developer platforms.
The skill instructs the agent to fetch performance thresholds and definitions from web.dev and developer.chrome.com (SKILL.md, Retrieval Sources).
It suggests that the user configure the chrome-devtools-mcp package via npx from the NPM registry to enable the skill's functionality (SKILL.md, Verify MCP Tools Available).

web-perf