The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through external user data.\n
Ingestion points: The 'Episode viewer promise' (a user-provided string) is ingested and passed directly into subagent prompts in SKILL.md.\n
Boundary markers: None; the instructions for spawning subagents do not include delimiters or protective instructions to ignore potential commands within the viewer promise.\n
Capability inventory: The skill utilizes Read, Write, Bash, and Task tools, providing a wide range of actions that could be manipulated if an injection occurs.\n
Sanitization: No validation or escaping logic is applied to the input before it is used in prompt construction.\n- [COMMAND_EXECUTION]: The skill uses the Bash tool to execute a local Python script helpers/thumbnail_render.py for rendering image previews. While the path is static, this involves executing system-level commands.

thumbnail-packager