How does the skill identify which prompts to optimize?

It utilizes Serena MCP to search for LLM clients and prompt construction points within the codebase; if unavailable, it performs a manual analysis using file system tools.

Will this skill change my LangGraph nodes or edges?

No, the skill is specifically designed to optimize the content and logic within existing nodes while strictly preserving the established graph structure and data flow.

Can this skill help with model cost reduction?

Yes, it incorporates optimization techniques such as temperature adjustment, model selection analysis, and prompt caching strategies to significantly reduce operational expenses.

What is the purpose of the .langgraph-master/fine-tune.md file?

This file acts as the primary configuration hub where optimization goals, evaluation criteria, and target metrics are defined to guide the iterative improvement process.

LangGraph Fine-Tuning Optimizer

Name: LangGraph Fine-Tuning Optimizer
Author: hiroshi75

byhiroshi75

•

データサイエンスとML

Optimizes LangGraph application performance through iterative prompt engineering and node-level logic refinements based on quantitative evaluation criteria.

概要

The LangGraph Fine-Tuning skill provides a systematic framework for improving the performance of existing agentic workflows by iteratively optimizing LLM prompts and processing logic within individual nodes. By utilizing Serena MCP for codebase analysis and maintaining strict graph structure integrity, it guides Claude through a rigorous cycle of baseline measurement, data-driven improvements, and re-evaluation. This skill is ideal for developers who need to enhance output quality, reduce error rates, or minimize token costs in established LangGraph applications without altering the underlying graph architecture or state schema.

主な機能

Iterative prompt and logic optimization for individual LangGraph nodes
Automated identification of LLM prompts using Serena MCP codebase analysis
Quantitative performance measurement via baseline and post-improvement evaluation
Strict preservation of existing graph structure, nodes, and edges
Implementation of advanced techniques including few-shot prompting and prompt caching
1 GitHub stars

ユースケース

Standardizing evaluation metrics and testing for agentic applications
Improving the accuracy and response quality of complex AI agent workflows
Reducing API costs and latency through prompt optimization and model selection

概要

主な機能

Iterative prompt and logic optimization for individual LangGraph nodes
Automated identification of LLM prompts using Serena MCP codebase analysis
Quantitative performance measurement via baseline and post-improvement evaluation
Strict preservation of existing graph structure, nodes, and edges
Implementation of advanced techniques including few-shot prompting and prompt caching
1 GitHub stars

ユースケース

Standardizing evaluation metrics and testing for agentic applications
Improving the accuracy and response quality of complex AI agent workflows
Reducing API costs and latency through prompt optimization and model selection