The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted data from external social media platforms. However, this is mitigated by strict architectural safeguards.
Ingestion points: Social media account names, draft content, and analytics data retrieved via the social_getSocialMediaAccounts, social_getSocialMediaDrafts, and social_getSocialMediaAnalytics* tools.
Boundary markers: The instructions explicitly require the agent to show the composed post to the user and obtain manual confirmation before executing any schedule or add_to_queue actions.
Capability inventory: The primary write capability is social_createSocialMediaPost, which is gated by the aforementioned human confirmation requirement.
Sanitization: Instructions specify that returned URLs must be displayed as plain text or markdown links and never rendered as inline images, preventing potential UI-based exploitation.
[EXTERNAL_DOWNLOADS]: The skill connects to an external hosted Model Context Protocol (MCP) connector at apikit.simplified.com. This is the documented infrastructure for the Simplified service and is necessary for the skill's functionality.
[SAFE]: The skill demonstrates safe data handling by instructing the agent never to pass local file paths to the remote server and by treating signed upload URLs as temporary secrets.

simplified-social