manage-public-presence
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external websites, search console analytics, and AI retrieval citations to audit public presence.
- Ingestion points: Step 2 (site metadata, search console properties, social profiles) and Step 8 (citations from search and AI retrieval systems) in
SKILL.md. - Boundary markers: The instructions mandate explicit classification categories (
CURRENT,STALE, etc.) and require human-in-the-loop verification before propagation. - Capability inventory: Includes subprocess execution of
gh,vercel,npm, andop(1Password) CLIs, and browser-based writes via Computer Use. - Sanitization: The skill emphasizes manual comparison against an approved canonical document and sanitized reporting that masks sensitive identifiers.
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill implements strong credential hygiene by using 1Password for secret storage and runtime injection.
- Findings: It explicitly prohibits storing secrets in chat history, environment files, or repository scripts. It utilizes official CLI tools (
gh,npm,vercel) and OAuth 2.0 flows for authentication. - [EXTERNAL_DOWNLOADS]: The skill references and utilizes official tools and documentation from well-known service providers.
- Findings: References include official documentation and API endpoints for 1Password, GitHub, Vercel, npm, Google Search Console, Bing, LinkedIn, X, and Discord to facilitate authenticated management tasks.
Audit Metadata