The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests creator data such as social media handles and campaign reasons. This represents a potential surface for indirect prompt injection; however, the risk is mitigated by the skill's core architecture which mandates human review for all generated outreach and drafts. Additionally, the UI implementation uses escapeHtml for sanitization, preventing XSS-based injection in the companion app.
[EXTERNAL_DOWNLOADS]: The skill utilizes the busabase-sdk and standard web framework dependencies (hono). These are well-known libraries required for the application's functionality and are sourced from official registries.
[CREDENTIALS_SAFE]: Security instructions within SKILL.md explicitly direct the agent not to request API keys or secrets from the user in chat. The system relies on standard environment variables and OAuth for backend communication, ensuring sensitive tokens are not exposed in the conversation history.
[COMMAND_EXECUTION]: The skill includes administrative scripts for executing marketing decisions. These scripts perform state updates within the Busabase database and do not execute arbitrary code from untrusted sources.

kelly-creators