dby-theme
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
curl,node, andjqto communicate with the Doubaoya API and process theme data. It specifically runs a bundled validation scriptscripts/validate-theme.mjsto verify theme configurations. - [EXTERNAL_DOWNLOADS]: The skill interacts with the vendor domain
https://doubaoya.comto fetch theme templates, render HTML, and synchronize user configurations. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted inputs, such as user-provided style descriptions and local Markdown files, creating a potential surface for indirect injection.
- Ingestion points: Data enters the context via user requests for theme modifications and the
sample.mdfile used for rendering previews. - Boundary markers: The skill includes instructions to run
scripts/validate-theme.mjsfor self-checking generated themes before submission, which acts as a validation boundary. - Capability inventory: The agent environment has capabilities for network requests via
curl, script execution vianode, and file manipulation usingjq,tee, andcp. - Sanitization: The included
scripts/validate-theme.mjsperforms a comprehensive safety scan of all generated theme strings, explicitly filtering out dangerous patterns such as<script>,<style>,class=,javascript:URIs, inline event handlers, anddata:URIs to prevent malicious injections into the WeChat editor environment.
Audit Metadata