ClaudeMap

Code with Claude Workshops

Anthropic's official "Code with Claude" workshop materials — nine hands-on courses on skills, MCP, evals and Claude Managed Agents.

Tutorials & Guidesofficialworkshopsanthropicmanaged-agentsevalstraining
Visit projectanthropics/cwc-workshops

Nine complete workshops Anthropic ran publicly: sweep an eval suite across models and inference parameters (rightmodel), decompose a 400-line-prompt agent into skills plus callable agents (agent-decomposition), reproduce Anthropic's own three-phase product workflow (how-we-claude-code), ship a first Claude Managed Agent through seven small API calls (ship-your-first-managed-agent), compete on agent configs (agent-battle), layer in memory primitives and a dreaming service (agents-that-remember), iterate a PPTX-generating agent through six measured variants (eval-driven-agent-development), coordinate a multi-agent M&A research team (production-ready-agent), and build an SEC-filings research desk with sub-agents and scorecards (research-desk).

What it does

  • Nine workshops, each a runnable folder: model/parameter sweeps, multi-agent decomposition, memory, eval-driven development, and production coordinator patterns
  • "How We Claude Code" documents Anthropic's own workflow — interview to spec, divergent HTML design explorations, then a Vite + React app whose DOM contract lets an agent or CI verify components at runtime
  • Eval-first teaching: the PPTX workshop scores six agent variants with a two-layer grader (programmatic .pptx XML metrics + LLM-as-judge on rendered slides)
  • Later workshops target Claude Managed Agents end to end — from a first seven-function agent to per-ticker analyst sub-agents with outcome-graded scorecards

When to use it

The fastest official path from "I use Claude Code" to "I architect multi-agent systems": pick the workshop matching your gap and work it hands-on. Also prime material for teams running internal Claude training.

Getting started

Nothing to install — clone the repo and open each workshop folder's README:

git clone https://github.com/anthropics/cwc-workshops
cd cwc-workshops/rightmodel   # or any of the nine

Things to know

Explicitly labeled "Not maintained and not accepting contributions" — treat it as archived courseware, not a living product (links and Managed Agents APIs will drift). Apache-2.0. Assumes existing comfort with Claude Code, skills and MCP.

Repository

anthropics/cwc-workshops

Charted

Related on the map

  • Building Effective Agents

    Guide

    Anthropic's canonical essay on agent design — workflows vs agents, and five composable patterns.

    • official
    • agents
    • engineering
    • patterns
    View

  • 12-Factor Agents

    Guide

    HumanLayer's 12 design principles for production-grade LLM agents — the closest thing to a 'best practices' manifesto.

    • agents
    • best-practices
    • production
    • principles
    View