The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines an audit command that processes untrusted data from external URLs and local source files to generate MD3 compliance reports.
Ingestion points: The audit command in SKILL.md instructs the agent to read external web content using mcp__claude-in-chrome__navigate and mcp__claude-in-chrome__read_page.
Boundary markers: The skill does not define specific delimiters or instructions to ignore potential injections within the audited content.
Capability inventory: The skill suggests using shell commands (grep) and browser automation tools, providing an execution surface for the agent.
Sanitization: No filtering or sanitization steps are documented for the content ingested during the audit process.
[EXTERNAL_DOWNLOADS]: The skill recommends the use of official Google Material Design libraries for web implementation.
Mentions @material/web and @material/material-color-utilities as standard dependencies.
These references target well-known, official packages from the official NPM registry and do not escalate the security risk.

material-3