The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data (such as blog posts, podcast transcripts, or video content) and transform it into new formats.
Ingestion points: Processes external content provided as input to the agent context.
Boundary markers: The provided examples do not use explicit delimiters or instructions to ignore embedded commands within the source content.
Capability inventory: The skill utilizes shell commands via the belt CLI tool to interact with external AI models (image generation, text-to-speech, etc.).
Sanitization: No explicit sanitization or filtering of input content is mentioned in the instructions.
[COMMAND_EXECUTION]: The skill provides examples of shell commands executed through the belt CLI.
Mitigation: The skill's YAML frontmatter includes allowed-tools: Bash(belt *), which restricts the shell execution environment to only allow commands starting with belt, significantly reducing the risk of arbitrary command injection.

content-repurposing