The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by instructing the agent to ingest external content and perform sensitive actions based on that data.
Ingestion points: The agent is directed to use search_web and browse_webpage to retrieve external documentation and read_file to inspect existing plugins and market metadata (referenced in SKILL.md).
Boundary markers: The instructions do not specify the use of delimiters or clear "ignore instructions" directives when the agent is processing data retrieved from the web or project files.
Capability inventory: The agent has a powerful set of capabilities, including writing and patching files (write_file, apply_patch), executing system commands (execute_command), and performing system-level updates via the moviepilot_api (such as config.system.update).
Sanitization: The skill does not provide instructions for validating or sanitizing the content fetched from the web before it is used to generate plugin code or execute system operations.

create-moviepilot-plugin