figma-use-slides
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided briefs and design guidelines to generate Figma slides, creating an indirect prompt injection surface.
- Ingestion points: User prompts, brand guidelines, and external Figma reference files are ingested as design inputs (documented in SKILL.md and references/slide-design.md).
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when handling external user-supplied data.
- Capability inventory: The skill uses the use_figma MCP tool to execute Plugin API code and the upload_assets tool for image management (found in SKILL.md and references/slide-content.md).
- Sanitization: No explicit sanitization or validation of user-controlled strings is described before they are incorporated into generated scripts or slide content.
- [DYNAMIC_EXECUTION]: The skill provides helper functions and templates to facilitate the generation of JavaScript code for the Figma Plugin API.
- Evidence: File references/slide-gotchas.md contains helper patterns like addFrame and addText that the agent is encouraged to use when constructing scripts for the use_figma tool. This is a low-risk, standard design for assisting with tool-based code generation.
- [EXTERNAL_DOWNLOADS]: The CHANGELOG.md file references a public repository on GitHub (github.com/figma/mcp-server-guide) as the upstream source for the workflow. This reference is documented neutrally as it points to a well-known technology service.
Audit Metadata