schedule-social-posts
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements strong safety best practices by requiring a confirmation step in Step 6 ("Confirm first") before any content is published publicly. It also utilizes a draft-first approach using the
create_posttool with thedraft: trueflag. The instructions specifically warn the agent not to invent claims, links, or statistics, which mitigates risk related to misinformation or malicious link generation.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided captions and external media URLs, creating a surface for potential indirect prompt injection. However, this is inherent to its primary purpose as a social media connector and is mitigated by human-in-the-loop requirements.\n - Ingestion points: User-supplied post content (SKILL.md, Step 4) and public media URLs (SKILL.md, Step 5).\n
- Boundary markers: Mandatory confirmation requirement in Step 6 ("Confirm before anything goes live") and account health checks in Step 2.\n
- Capability inventory: Tools including
list_accounts,upload_media_from_url,create_post, andschedule_post.\n - Sanitization: Instructions in Step 4 explicitly prevent the agent from inventing links, statistics, or claims.
Audit Metadata