skills/redongreen/uspec/extract-color/Gen Agent Trust Hub

extract-color

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill employs dynamic code generation for its Step 3-delta feature. It can construct and execute small JavaScript snippets (under 50 lines) using Figma MCP tools like figma_execute or use_figma to resolve missing metadata. The instructions restrict these scripts to read-only operations and require every call to be recorded in the output metadata for auditing purposes.- [PROMPT_INJECTION]: The skill presents an indirect prompt injection attack surface because it ingests and reasoning upon data from multiple external sources.
  • Ingestion points: Data is loaded from _base.json, api-dictionary.json, and the user-supplied optionalContext parameter.
  • Boundary markers: There are no specific delimiters or instructions to the model to ignore potential instructions embedded within the processed data.
  • Capability inventory: The skill has capabilities to write files to the workspace and execute code within the Figma environment via MCP adapters.
  • Sanitization: The skill workflow does not mention any validation, escaping, or sanitization of the input data before it is interpolated into the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 10:11 PM
Security Audit — agent-trust-hub — extract-color