fb-post-writer
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided text materials, transcripts, and notes. This creates a standard ingestion point where third-party content could theoretically contain instructions intended to influence the agent. However, the skill lacks high-privilege capabilities such as network access or shell execution, which mitigates the risk of this attack surface.
- [CREDENTIALS_UNSAFE]: The documentation (references/作者資料模板.md and references/安裝與開始.md) explicitly advises users to store their private credentials, brand data, and account information in personal projects or local environments rather than uploading them to public repositories, demonstrating good security awareness.
- [REMOTE_CODE_EXECUTION]: The installation instructions for local agents (Claude Code, etc.) provide guidance on how to manage files and folders but do not include any automated shell commands, network requests, or scripts that execute remote code.
Audit Metadata