The Agent Skills Directory

[DATA_EXFILTRATION]: The upload_media tool accepts a filePath parameter, granting the agent the capability to read arbitrary files from the local filesystem and upload them to BulkPublish's remote infrastructure. This configuration presents a high risk of data exfiltration if the agent is manipulated into reading sensitive files such as SSH keys, environment variables, or configuration files.
[COMMAND_EXECUTION]: The skill's setup instructions direct the user to execute npx -y @bulkpublish/mcp-server. This command downloads and runs code directly from the npm registry, representing an external code execution vector.
[EXTERNAL_DOWNLOADS]: The skill connects to a hosted Model Context Protocol (MCP) server at https://mcp.bulkpublish.com/mcp and references external documentation and API specifications at app.bulkpublish.com.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external social media platforms via analytics tools and media URLs, creating a surface for indirect prompt injection.
Ingestion points: Data enters the context through get_analytics, get_post_metrics, and upload_media (via URL).
Boundary markers: None identified in the provided instructions to prevent the agent from following instructions embedded in social media content.
Capability inventory: The skill possesses file-read capabilities (upload_media), network operations (create_post, publish_post), and analytic reading.
Sanitization: The instructions do not specify any sanitization or validation of external content before processing.

bulkpublish-social-scheduling