AI agent dev stack (LangGraph + LangSmith)

Install a skill or read the field note below to see how we apply this pattern in real Claude Code projects.

verified today5 min
Install the workflow

For building, debugging, and evaluating multi-step LLM agents. The skills that catch the problems traces-first: cache misses, regression in eval scores, trace forensics. Run this command to install a skill and start from a working baseline instead of rebuilding the setup from scratch.

$npx frenxt-cables add stack-ai-agent
Did install work?
Files this command writes (4 files)
  • .claude/skills/stack-ai-agent/SKILL.md artifact/SKILL.md
  • .claude/skills/analyse-langsmith-trace/SKILL.md artifact/skills/analyse-langsmith-trace/SKILL.md
  • .claude/skills/debug-prompt-caching/SKILL.md artifact/skills/debug-prompt-caching/SKILL.md
  • .claude/skills/benchmark/SKILL.md artifact/skills/benchmark/SKILL.md

AI agent dev stack installed. Pair with `stack-fullstack` if the agent is part of a full product surface.

AI agent dev stack (LangGraph + LangSmith)

For building, debugging, and evaluating multi-step LLM agents. The skills that catch the problems traces-first: cache misses, regression in eval scores, trace forensics.

One line

npx -y frenxt-cables stack stack-ai-agent --force

What it installs

Claude plugins (4):

  • superpowers — systematic debugging, TDD, verification discipline
  • claude-md-management — keep the agent project's CLAUDE.md current
  • vercel — AI SDK, AI Gateway routing, Workflow DevKit for durable runs, Sandbox for untrusted code
  • sentry — AI agent monitoring (LLM call tracing, token cost, latency)

Skills:

  • analyse-langsmith-trace — paste a LangSmith URL, get a structured triage
  • debug-prompt-caching — forensics on cache misses and cost spikes
  • benchmark — run LangGraph benchmarks, read traces manually

When to use this

New agent project where traces and evals will be your day-two tools. Works alongside stack-fullstack if the agent is part of a larger product.

When not

Pure UI work → stack-ux. Full product surface → layer stack-fullstack on top.

Quick answers

What do I get from this cable?

You get a skill plus a dated field note that explains how we use it in real Claude Code workflows.

How much time should I budget?

Typical effort is 5 min. The cable is marked intermediate.

How do I install the artifact?

Run npx frenxt-cables add stack-ai-agent. The install block shows the files it writes and any prerequisites before you run it.

How fresh is the guidance?

The cable is explicitly last verified on 2026-04-18, and includes source links for traceability.

Work with FRE|Nxt

We build the production AI systems we write about.

Cables are the field notes. The playbooks come from client engagements — multi-agent systems, RAG pipelines, and LLM cost cuts that ship and hold up in production. If something here maps to a problem on your roadmap, two ways in:

Audit capacity: 5 slots/month · No pitch deck · NDA on request

Same shelf · Fix a specific problem
Share this cable