designcraft
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides detailed instructions for driving the DesignCraft application through its primary automation surfaces:
- A command-line interface (designcraft-cli) for headless document processing and export.
- An MCP (Model Context Protocol) server for tool-based agent interaction.
- A local TCP control channel (default port 7979) using JSON-RPC for driving the live desktop application.
- [INDIRECT_PROMPT_INJECTION]: The skill processes various external file formats which could potentially contain malicious instructions meant to influence the agent's behavior during document processing.
- Ingestion points: Untrusted data enters the context when the agent opens or places files in formats such as .idml, .docx, .rtf, .md, .xlsx, .csv, or .tsv via commands like file.open, file.place, or data.source.select.
- Boundary markers: There are no explicit instructions to use delimiters or ignore embedded prompts within the ingested content.
- Capability inventory: The skill possesses the capability to write files to the local file system (file.exportPdf, file.save), execute a wide range of application commands (engine.execute), and perform UI actions (ui.click, ui.drag).
- Sanitization: The documentation does not describe any specific sanitization or validation steps for content extracted from these external formats.
Audit Metadata