llm-generation

Implement and trace application-level LLM generation workflows with structured output and versioning.

82.1k|15.8k|Updated May 21, 2023

Search Tags:#tracing #prompt-engineering #structured-output #lLM #AI-generation #Observability

One-click install

npx skills add https://github.com/lobehub/lobehub --skill llm-generation

Or copy as Structured Prompt for Agent▼

Please help me install this Agent Skill.
Skill: llm-generation
Source: https://github.com/lobehub/lobehub/tree/main/.agents/skills/llm-generation
Command: npx skills add https://github.com/lobehub/lobehub --skill llm-generation

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill addresses the fragmentation and lack of observability in LLM-based application features, ensuring that prompts, models, and structured outputs are managed as stable, testable, and traceable workflows.

Core Features & Use Cases

Structured Generation: Enforces schema-based output validation and consistent naming conventions for LLM responses.
Observability & Tracing: Integrates with tracing services to monitor latency, cost, and quality across different prompt versions and scenarios.
Use Case: When building a new AI-driven content drafting feature, use this Skill to define the system prompt, version the schema, and configure the tracing scenario to ensure consistent performance and auditability.

Quick Start

Use the llm-generation skill to implement a new structured generation workflow with defined prompt versions and tracing scenarios.

Frequently Asked Questions about llm-generation

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I version prompts for LLM generation?▼

Store prompt versions as 'v<major>' or 'v<major>.<minor>' strings alongside the prompt content. Export both the prompt and the version constant from the same module to ensure they remain synchronized during updates.

What is the purpose of a tracing scenario?▼

A tracing scenario represents a stable product workflow and lifecycle stage. It allows for accurate tracking of latency, cost, and success rates by isolating data for specific business actions rather than individual prompts.

How should I handle structured output schemas?▼

Assign each JSON schema a stable, workflow-appropriate name and validate the generated content at the service boundary. Ensure prompt instructions and schema requirements are strictly aligned to prevent validation errors.

When should I create a new tracing scenario?▼

Create a new scenario whenever the business action differs, even if the prompt or schema is reused. This prevents contamination of performance metrics across different user-visible workflows.