The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, which could be used to influence the agent's behavior.
Ingestion points: The skill processes external "product feed reviews" and "tabular product datasets" (SKILL.md, Security Guardrails ASI01).
Boundary markers: The instructions do not specify any delimiters or explicit safety instructions to isolate external data from the system prompt.
Capability inventory: The skill utilizes execute_command for CLI pipelines and write_file for content deployment.
Sanitization: Although rule ASI05 mentions the need for parameterization, the skill does not implement specific sanitization routines for incoming data.
[COMMAND_EXECUTION]: The skill uses the execute_command tool and references external CLI tools like FFmpeg, Remotion, and CapCut for video synthesis. If input parameters from product feeds are not properly handled, this could lead to command injection.
[EXTERNAL_DOWNLOADS]: The skill interacts with well-known AI and media service providers, including OpenAI, Anthropic, Google (Gemini), and ElevenLabs, to fetch generated content and audio assets.

generate-mmo-content