Skills
Navigate
skills/daangn/seed-design/seed-verify-figma-mcp-transports/Gen Agent Trust Hub

seed-verify-figma-mcp-transports

Pass

Audited by Gen Agent Trust Hub on Oct 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill prompts the user to provide a Figma layer URL. This untrusted external input is parsed to extract file keys and node IDs, which are then used as arguments in shell commands. This creates a surface for indirect prompt injection if the input is malicious.
  • Ingestion points: User-provided layer URL via the AskUserQuestion tool.
  • Boundary markers: Absent; the skill relies on the parseFigmaUrl logic for extraction without explicit boundary delimiters in the prompt.
  • Capability inventory: The skill has the ability to execute shell commands via bun and interact with the Figma API.
  • Sanitization: Uses a script-based helper (parseFigmaUrl) to process the input, but the verification of the extracted IDs before shell interpolation is not specified.
  • [COMMAND_EXECUTION]: The skill routinely executes local shell commands using the bun runtime to build project components and run testing probes. Examples include bun packages/mcp/scripts/probe-transports.ts and building the plugin via bun --filter figma-mcp build. These scripts are located within the vendor's local repository structure.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 11, 2026, 10:15 PM