The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is instructed to 'inspect the project (routes, screens, package manifest)' to infer application scope and determine which UX flow skills to apply. While intended for context gathering, this involves reading project structure and metadata.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface during the generation of the flow-report.html file.
Ingestion points: Project-level metadata (route names, screen titles, package manifests) and user-provided descriptions of the application features.
Boundary markers: No explicit boundary markers or 'ignore' instructions are used when the agent processes these project files to categorize them.
Capability inventory: The skill performs local file reads (project files and skill definitions) and file writes (emits flow-report.html to the project root).
Sanitization: There is no mention of sanitizing or escaping content derived from the project files or user input before it is interpolated into the report-template.html. This creates a surface where malicious strings in project files could lead to content injection or Cross-Site Scripting (XSS) within the generated HTML report.

userflow