The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands via the listenhub CLI tool to create slides, manage configuration, and handle audio downloads. These commands are parameterised by user input (e.g., --query "{topic}", --source-url "{url}") and configuration values. The skill uses jq for JSON processing and curl for downloading audio files.
[EXTERNAL_DOWNLOADS]: The skill downloads generated audio files using curl from listenhub.ai infrastructure ({audioUrl}). This is a core part of its functionality when narration is enabled.
[INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for untrusted data as it accepts topics, text content, and URLs from the user to generate slides. These inputs are interpolated into shell commands.
Ingestion points: Topic/Content input in Step 1 (SKILL.md).
Boundary markers: The skill uses double quotes for arguments (e.g., "{topic}") and relies on the CLI's internal handling of the query, but does not explicitly sanitize the input before interpolation.
Capability inventory: Executes shell commands (listenhub, curl, jq, mkdir, echo, cat) and writes files to the local directory.
Sanitization: None explicitly mentioned in the skill instructions; it relies on the underlying platform's shell execution safety and the CLI tool's own handling.

slides