figma-agent

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requires the execution of the figma-mcp-server via bunx, involving a shell command to host the necessary local API bridge.
  • [EXTERNAL_DOWNLOADS]: The documentation suggests that users install a complementary skill (design-diff) from the author's GitHub repository at https://github.com/planetabhi/design-diff.
  • [PROMPT_INJECTION]: The skill provides a surface for indirect injection by ingesting data from Figma files (node names, labels, and comments) that could contain malicious instructions.
  • Ingestion points: Design nodes, component metadata, and comments fetched via Figma API tools in SKILL.md and references/design-review.md.
  • Boundary markers: Absent; there are no specific instructions to distinguish design data from executable prompts or to ignore instructions embedded in layers.
  • Capability inventory: Shell command execution (for MCP server requirements), network operations to the Figma API, and file writing for generated code and documentation.
  • Sanitization: Absent; the skill does not specify methods to sanitize or validate ingested design content before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 08:10 PM
Security Audit — agent-trust-hub — figma-agent