The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes functionality to ingest external web content for trend analysis, which presents a surface for indirect prompt injection.
Ingestion points: The 'Trends' flow in references/trends-to-posts.md uses web tools to scan for industry news, news drops, and innovation stories.
Boundary markers: While specific delimiters for external content are not explicitly defined, the workflow requires the agent to pitch identified candidates to the user for approval before any drafting occurs.
Capability inventory: The skill utilizes tools to create and schedule social media posts (create_post, schedule_post, update_post).
Sanitization: The instructions mandate the use of validate_post_content for each platform before presenting drafts to the user, providing a secondary layer of validation.
[SAFE]: The skill correctly scopes sensitive operations to the postey hub, using vendor-specific MCP resources (postey://) for account management. All scheduling and publishing actions require explicit user consent and approval of draft content, preventing autonomous data exfiltration or unauthorized posting.

postey-ideas