The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides examples for Retrieval-Augmented Generation (RAG) where untrusted data from web searches is processed by an LLM.
Ingestion points: Content retrieved from external search tools like tavily/search-assistant and exa/search (SKILL.md).
Boundary markers: The provided bash examples interpolate variables like $SEARCH_RESULT and $CONTENT directly into prompts without using XML tags, triple backticks, or other delimiters to isolate untrusted content.
Capability inventory: The skill uses the belt CLI tool which has the capability to run various external applications and perform network operations.
Sanitization: The examples do not demonstrate sanitization or validation of the retrieved web content before it is passed to the LLM.
[EXTERNAL_DOWNLOADS]: The skill documentation references external skill packages and installation scripts from the tool provider's official repositories.
Evidence: Installation instructions point to npx skills add belt-sh/cli and https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md (SKILL.md).
[COMMAND_EXECUTION]: The skill's primary functionality relies on executing commands via the belt CLI to manage remote applications and inference tasks.
Evidence: Frequent usage of belt app run for search, extraction, and LLM invocation throughout the pipeline examples in SKILL.md.

ai-rag-pipeline