The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines workflows that ingest external data (such as blog content) and pipe the output of one AI model into another. This creates a potential surface for indirect prompt injection where instructions hidden in the input data could influence subsequent automated steps.
Ingestion points: Processes external content and LLM-generated scripts as input for the belt CLI (SKILL.md).
Boundary markers: The workflows use structured JSON objects (--input '{\"prompt\": \"...\"}') to provide context to the tools, which helps separate data from instructions.
Capability inventory: The skill utilizes the belt CLI tool to perform remote inference, media merging, and file operations.
Sanitization: Input is passed as strings within JSON structures to the underlying models via the CLI tool.
[EXTERNAL_DOWNLOADS]: The skill references external dependencies and installation scripts to enable its functionality.
The instructions suggest installing belt-sh/cli using a platform-specific package manager (npx skills add).
It links to installation documentation hosted on GitHub (raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md) for the required CLI utility.

ai-content-pipeline