Skip to content
Kitbay
Skill

llm-generation

NOASSERTIONUpdated on Nov 11, 202684,974GitHub stars

A contributor skill in the LobeHub repository for adding or changing language model calls in the product, such as application prompts and structured or text generation.

What the skill teaches

A contributor skill in the LobeHub repository for adding or changing language model calls in the product, such as application prompts and structured or text generation. It points the agent to the packages that hold reusable prompts, the server generation wrapper and the tracing scenario list, and sets rules: each prompt exports its message builder, JSON schema, schema name and version together; a tracing scenario names one user-facing workflow stage and is never borrowed from a nearby one; models come from the owning service's configuration; and output is validated at the service boundary. A checklist asks for tests on emitted tracing fields and validation failures, then the repository's type check.

Capabilities

  • Place prompts and schemas in LobeHub's prompts package
  • Version prompts and choose tracing scenarios
  • Validate structured output at the service boundary