The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions and tools process user-supplied input strings for image and video generation tasks, creating a potential surface for indirect prompt injection.
Ingestion points: The prompt parameter used in generation tasks (e.g., imagine, edit, blend, videos) across the SKILL.md examples and MCP tool definitions.
Boundary markers: The skill does not specify explicit boundary markers or delimiters to separate untrusted prompt data from agent instructions.
Capability inventory: The skill utilizes network operations via curl to interact with external API endpoints and provides access to several Midjourney-related MCP tools.
Sanitization: No explicit sanitization or filtering of user prompts is described in the skill configuration.
[EXTERNAL_DOWNLOADS]: The skill documentation describes the installation of software packages and interaction with remote services managed by the author.
References the installation of the @acedatacloud/midjourney and mcp-midjourney packages.
Interacts with the api.acedata.cloud API and a hosted MCP server at midjourney.mcp.acedata.cloud for image and video processing tasks.

midjourney-image