The Agent Skills Directory

[SAFE]: The skill connects to a hosted MCP connector at https://apikit.simplified.com/mcp. This is a legitimate service endpoint for the Simplified.com platform.
[SAFE]: Authentication is performed via OAuth, which avoids the use of hardcoded API keys or other sensitive credentials.
[SAFE]: The instructions include clear security warnings against exposing client-local file paths to the remote server and correctly advise treating temporary signed upload URLs as secrets.
[INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting external data such as account names and post messages which could be attacker-controlled. Ingestion points: account metadata from social_getSocialMediaAccounts and post messages from social_getSocialMediaAnalyticsPosts. Boundary markers: None specified. Capability inventory: allows writing content to external social platforms via social_createSocialMediaPost. Sanitization: No explicit instructions for sanitizing retrieved content. The risk remains low as this behavior is core to the social media management functionality.

simplified-social