LLM Application Dev
Bundles the RAG, vector-index and prompting skills an LLM app needs, rebuilt around LangGraph.
Install
Ships scriptsClaude Code: /plugin marketplace add wshobson/agents → /plugin install llm-application-dev@claude-code-workflows
RAG, vektör arama ya da LangGraph tabanlı bir ajan kurulurken; istem kalıpları ölçülerek iyileştirilirken
- Author
- Seth Hobson
- License
- MIT
Eight skills and three agents cover the parts of an LLM application that are easy to get wrong: chunking and reranking for retrieval, HNSW and IVF index parameters, hybrid fusion of vector and keyword scores, and prompt patterns with a Python script that rewrites a prompt against measurable criteria. The 2.0 rewrite dropped LangChain's deprecated initialize_agent() for LangGraph StateGraph with checkpointers, which is the difference between an agent that resumes after a crash and one that starts over. Embedding advice is opinionated rather than neutral: Voyage models are the default recommendation for Claude-based stacks. A separate evaluation skill exists so retrieval changes get measured instead of eyeballed.
Similar skills
DeepEval Skills
SkillEvaluation, tracing and OpenTelemetry export for LLM apps, split into three non-overlapping skills.
RAG Implementation
SkillBuilds retrieval pipelines that ground answers in real documents.
LLM Finetuning
SkillRuns fine-tuning as a gated lifecycle where the eval harness is built before any training config.
Comments(0)
Sign in to comment