The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references installation scripts and related skills from inference.sh and its associated GitHub repositories. These resources are integral to the skill's stated functionality and originate from the vendor's infrastructure.
[COMMAND_EXECUTION]: The skill employs the belt CLI tool to perform various automated tasks, including capturing screenshots via an agent-controlled browser and generating product visuals using AI models. Command execution is appropriately restricted to the belt utility through the allowed-tools frontmatter configuration.
[PROMPT_INJECTION]: The skill handles untrusted data provided by users (such as raw changelog details) to produce formatted updates, presenting an indirect prompt injection surface.
Ingestion points: User-provided changelog and release note content (SKILL.md).
Boundary markers: The instructions do not define delimiters or specific safety instructions for the agent when processing external text.
Capability inventory: Subprocess execution via Bash(belt *) providing access to browser automation and image generation tools (SKILL.md).
Sanitization: No explicit validation or filtering of user-supplied content is described.

product-changelog