The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external content (such as blog posts, podcasts, and videos) to generate derivative assets, creating a vulnerability surface for indirect prompt injection where instructions hidden in source material could influence the agent.
Ingestion points: Source material is ingested for conversion into threads, scripts, and posts throughout SKILL.md.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the source content.
Capability inventory: The skill utilizes the belt CLI to perform sensitive actions including social media posting (x/post-create), image generation, and text-to-speech, as documented in SKILL.md.
Sanitization: No explicit sanitization or validation of the input content is described before it is passed to AI models or tools.
[COMMAND_EXECUTION]: The skill makes extensive use of the belt CLI tool via Bash to interact with remote models and execute tasks like image generation and social media automation.
[EXTERNAL_DOWNLOADS]: The skill instructions reference external resources and installation scripts hosted on GitHub (raw.githubusercontent.com/inference-sh/...).
[DYNAMIC_EXECUTION]: The skill provides shell script examples, including loops, to dynamically generate and execute commands for batch processing assets like presentation slides.

content-repurposing