The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides an audit command that instructs the agent to analyze external content from URLs or local file paths. This creates a potential surface for indirect prompt injection if the targeted content contains malicious instructions designed to influence the agent's auditing logic.
Ingestion points: The audit command in SKILL.md accepts a URL or file path as input for analysis.
Boundary markers: The instructions do not specify any delimiters or safety prompts to isolate external content from the agent's core instructions during the audit process.
Capability inventory: The skill leverages browser tools for inspecting live URLs and file system access for reading source code to generate compliance reports.
Sanitization: There are no explicit instructions for the agent to sanitize or escape data retrieved from external sources before processing.
[COMMAND_EXECUTION]: The skill's SKILL.md file suggests using standard shell commands like grep to audit local project files for hardcoded color values and legacy Material Design 2 imports. These commands are intended for static analysis of the user's codebase.
[EXTERNAL_DOWNLOADS]: The documentation references the installation of official Material Design packages from the @material scope via npm, which are standard utilities for theme generation and component usage on the web.

material-3