The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data such as blog posts, podcast transcripts, and videos to generate derivative social media content. This functionality presents a risk of indirect prompt injection where malicious instructions hidden in the source material could be executed by the agent or influence its output.
Ingestion points: The skill instructions prompt the agent to read and "repurpose" long-form source content like blogs and transcripts provided in the context or via files (SKILL.md).
Boundary markers: There are no explicit delimiters or instructions provided to separate user-provided content from agent instructions, nor are there warnings to ignore embedded instructions in the source text.
Capability inventory: The skill uses the belt CLI tool to perform high-impact actions, including generating images, audio, video, and posting directly to social media platforms via belt app run x/post-create (SKILL.md).
Sanitization: No sanitization, validation, or filtering mechanisms are defined for the ingested external content.
[EXTERNAL_DOWNLOADS]: The skill encourages the installation of external CLI tools and additional skill packages from third-party repositories.
Evidence: The documentation suggests installing the belt-sh/cli and various other skills using npx skills add and provides links to external installation instructions hosted on GitHub at https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md (SKILL.md).
[COMMAND_EXECUTION]: The skill relies on shell command execution to perform its primary functions, including credential management and application execution.
Evidence: Includes multiple bash examples such as belt login for authentication and belt app run for interacting with remote AI models and social media APIs (SKILL.md).

content-repurposing