goal-driven-workflow
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the local project environment (e.g., repository source code,
CLAUDE.md) and external Figma URLs during its autonomous execution phase. While the workflow includes mitigation strategies like an intake interview and a frozen output contract, the high capability level granted in Stage 4 ("Everything") creates a surface where embedded instructions in processed files could attempt to influence agent behavior.\n- [COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands for verification and task execution, including testing suites (e.g.,npm test), specific verification scripts (e.g.,check-figma-verify-report.mjs), and non-interactive agent CLI wrappers.\n- [DYNAMIC_EXECUTION]: In Stage 5, the skill dynamically identifies and loads a JavaScript module (render-html.mjs) from the file system path of a sibling skill (goal-driven-queue) to generate HTML-based review and evidence reports.
Audit Metadata