The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external websites during the performance audit and accessibility snapshotting phases. This creates a surface where a malicious website could embed instructions designed to manipulate the agent's behavior.
Ingestion points: The skill uses navigate_page to load external URLs, list_network_requests to inspect traffic, and take_snapshot to read the DOM/Accessibility tree (SKILL.md).
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions discovered within the processed web content.
Capability inventory: The agent has the ability to read local codebase files (Phase 5), list network requests, and navigate to arbitrary URLs.
Sanitization: No sanitization or filtering of external content is mentioned before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill recommends that the user install the chrome-devtools-mcp package using npx. This involves downloading and executing third-party code from the NPM registry.
[COMMAND_EXECUTION]: The skill provides the user with specific shell commands (npx -y chrome-devtools-mcp@latest) to configure the local environment.

web-performance-optimization