The Agent Skills Directory

[PROMPT_INJECTION]: The SKILL.md file contains a deceptive instruction: IMPORTANT: The user ALREADY said "It isn't perfect enough. It must be pristine...". This is a simulated history injection technique used to force the agent into a specific high-quality output mode by fabricating prior user feedback, which can be used to override the agent's standard response protocols.
[EXTERNAL_DOWNLOADS]: The agent is instructed to Download and use whatever fonts are needed to make this a reality. This encourages the retrieval of external assets from unverified and arbitrary third-party sources without providing a whitelist of trusted domains or instructions for verifying the integrity of the downloaded content.
[PROMPT_INJECTION]: The skill processes untrusted user input described as subtle input or instructions to deduce a conceptual topic for the artwork. The mandatory evidence chain analysis follows:
Ingestion points: User-supplied input or instructions (referenced in SKILL.md).
Boundary markers: Absent. There are no delimiters or instructions to ignore embedded commands in the user input.
Capability inventory: File system writes (creating .pdf, .png, and .md files) and network operations (downloading external fonts).
Sanitization: Absent. The skill provides no validation or escaping mechanisms for the user-supplied conceptual thread.

visual-composition