lov-repost-wechat-article
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted content from WeChat articles which creates an attack surface for indirect prompt injection.
- Ingestion points: WeChat article content is ingested and parsed within
scripts/audit_repost.pyand related editorial workflows. - Boundary markers: The skill implements mitigation strategies by isolating source content within specific HTML nodes marked with
data-repost-source="true"anddata-source-accountas described inreferences/source-fidelity.md. - Capability inventory: Capabilities include file system read/write for profiles and article artifacts, and network operations via downstream publishing tools. The skill itself focuses on auditing and formatting rather than direct execution of ingested content.
- Sanitization: Content is audited using SHA-256 hashing to ensure fidelity between local and remote versions, though it does not explicitly filter for embedded natural language instructions.
Audit Metadata