The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external transcript data and project-specific notes, which could contain malicious instructions.
Ingestion points: Reads video transcripts and AGENTS.md to identify topic spans and build coverage maps.
Boundary markers: The skill does not define clear delimiters or instructions to ignore potential commands embedded within the transcript text.
Capability inventory: The skill has access to shell commands (rg, Select-String), Node.js and PowerShell script execution, and package execution via npx.
Sanitization: There is no evidence of sanitization or input validation for transcript content before it is used to drive automated logic.
[COMMAND_EXECUTION]: The skill explicitly instructs the agent to perform file system searches and project validation using shell utilities.
Evidence: Directives to use rg (ripgrep) and Select-String for inspecting root compositions and transcript chunks, as well as npx hyperframes lint for project validation.
[DYNAMIC_EXECUTION]: The workflow involves the use of scripts to process data at runtime.
Evidence: The instructions suggest using "small Node/PowerShell transcript scripts" to handle transcript extraction and processing, which implies the execution of dynamically generated or locally stored script logic.

hyperframes-video-beats