suede-instagram-growth
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, creating a surface for potential indirect prompt injection.
- Ingestion points: The skill reads user-supplied content ledgers, post URLs, CRM data, audience comments, and DMs, as detailed in
references/account-audit.mdandreferences/daily-loop.md. - Boundary markers: The instructions include a comprehensive "Boundaries" section and a "HALT" contract framework to prevent unauthorized actions.
- Capability inventory: The skill can generate content (Reels scripts, carousel copy, captions) and provides instructions for publishing content if authorized by the user.
- Sanitization: Instructions in
references/daily-loop.mdmandate that comments or DMs containing threats, private data, or security issues must be escalated for human review instead of being processed automatically. - [EXTERNAL_DOWNLOADS]: The skill contains references to external documentation and API resources.
- Evidence: Links to official Meta/Facebook help pages and the official Instagram API collection on Postman are provided in
references/account-audit.md. These are well-known and trusted technology services.
Audit Metadata