The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs several system-level operations to manage video libraries and exports. It executes local Ruby scripts located in the lib/ directory (e.g., lib/buttercut/library.rb, lib/buttercut/export.rb, lib/buttercut/contact_sheet.rb) and uses shell commands such as cp to copy files to the Desktop and open to launch professional editing software. These actions are consistent with the skill's primary function.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its core logic involves a sub-agent reading and interpreting video transcripts (transcripts/*.json) and footage summaries (summaries/*.md). If these files—which are derived from external video content—contain malicious instructions, they could influence the agent's behavior during the roughcut generation process.
Ingestion points: The sub-agent reads transcript JSON files and markdown summaries from the library directory.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are used when interpolating transcript data into the prompt.
Capability inventory: The skill can execute shell commands, perform file system operations (read/write/copy), and invoke other sub-agents.
Sanitization: There is no explicit sanitization or validation logic described for the content of the transcripts before they are processed by the LLM.

cut