figma-create-design-system-rules

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection. It ingests data from an external Figma MCP server (via tools like get_design_context) and incorporates it into generated rule files and code implementations. A malicious Figma file could potentially embed instructions to subvert the agent's behavior during code generation.
  • Ingestion points: Figma MCP server outputs (SKILL.md Step 3 and 4).
  • Boundary markers: No explicit delimiters or warnings are defined to instruct the agent to ignore instructions embedded within the Figma design data.
  • Capability inventory: The skill allows writing to project configuration files such as CLAUDE.md, AGENTS.md, and .cursor/rules/ (SKILL.md Step 4).
  • Sanitization: There are no instructions for sanitizing or filtering the content retrieved from the Figma API before it is processed.
  • [COMMAND_EXECUTION]: The skill facilitates the creation and modification of project-level rule files. While this is the core function of the skill, the ability to modify agent instructions (like .cursor/rules) based on external inputs constitutes a significant capability that should be monitored for misuse.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 09:08 AM
Security Audit — agent-trust-hub — figma-create-design-system-rules