The Agent Skills Directory

[DATA_EXFILTRATION]: The skill reads data from files within the outputs/ and context/ directories (such as brand style guidelines and generated content) and transmits this information to the external Blotato service (blotato.com) to facilitate visual generation and social media scheduling.
[COMMAND_EXECUTION]: The skill makes extensive use of Blotato MCP tools to perform automated actions on external social media accounts, including blotato_create_post and blotato_create_visual.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes content from external files in the outputs/ directory without applying sanitization or boundary markers. This could allow maliciously crafted content from previous pipeline steps to influence the publisher's actions.
Ingestion points: Reads from outputs/linkedin/, outputs/threads/, outputs/x/, and outputs/captions/ directories.
Boundary markers: Absent. Content is used directly for visual generation and captions.
Capability inventory: Includes network-enabled social media posting and visual generation via the Blotato MCP server.
Sanitization: Absent. The skill does not validate or escape the content before passing it to the scheduling tools.

publisher