The Agent Skills Directory

Command Execution Considerations: The skill uses PowerShell and Bash to invoke system utilities like expand.exe and cabextract for extracting .xsn archives. This is a functional requirement for the migration, but it involves executing shell commands with parameters derived from user-provided arguments.
Indirect Prompt Injection Surface: The migration process involves using an Agent to read and inventory the contents of external InfoPath files (including manifest.xsf and schema1.xsd). As these files are processed by the agent to generate instructions for the downstream generate-canvas-app skill, there is a potential surface for indirect prompt injection if the source files contain content designed to influence the agent's behavior.
Ingestion points: manifest.xsf, schema1.xsd, template.xml, and XSL views are read from the extracted .xsn archive (SKILL.md).
Boundary markers: The skill does not explicitly define delimiters or instructions to ignore embedded commands during the parsing phase.
Capability inventory: The skill environment includes Bash, PowerShell, Write, Agent, and Skill tools (SKILL.md).
Sanitization: While the skill decodes URL-encoded field names, it does not specify general content sanitization for the XML/XSF data before processing by the agent.

infopath-to-canvas