design-html
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides an optional workflow to download the impeccable design detector engine from Paul Bakaus's GitHub repository. This process is guarded by an
AskUserQuestionand requires explicit user consent before any action is taken. - [EXTERNAL_DOWNLOADS]: Fetches the Pretext layout library from the well-known
esm.shCDN as a fallback if the local vendor bundle is missing. - [COMMAND_EXECUTION]: Starts a local HTTP server using
python3 -m http.serverbound to the loopback interface (127.0.0.1) to facilitate live preview of the generated design. - [COMMAND_EXECUTION]: Executes several vendor-provided binaries and shell scripts (e.g.,
gstack-design-approved,gstack-render.ts) to perform specialized design tasks and manage project state. These resources originate from the skill's author (garrytan). - [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface by processing untrusted data from product plans and design metadata.
- Ingestion points: Reads
CEO_PLANmarkdown files andapproved.jsondesign metadata (SKILL.md Step 0). - Boundary markers: The skill does not explicitly use boundary markers or "ignore embedded instructions" warnings when interpolating untrusted data into prompts.
- Capability inventory: The skill has extensive capabilities including shell command execution (
Bash), file creation (Write), and file modification (Edit). - Sanitization: No specific sanitization or filtering of external content is described before it is processed by the AI.
Audit Metadata