for people who will write the code

From Zero to Agent

Learn Python, backend systems and agent design through worked examples. Try supported Python snippets in the browser, then follow the chapters to build real integrations locally.

Start reading Open the agent lab

Optional: try a small code example

This is real Python syntax, parsed and executed by the TypeScript runtime with a simulated OpenAI-compatible SDK.

python · openai adapter

from openai import OpenAI

client = OpenAI(base_url="http://localhost:1234/v1", api_key="lm-studio")

reply = client.chat.completions.create(
    model="local-model",
    messages=[
        {"role": "system", "content": "You answer in one sentence."},
        {"role": "user", "content": "Why does an agent need tools?"},
    ],
)

print(reply.choices[0].message.content)
print("tokens:", reply.usage.total_tokens)

Browser interpreter: a limited Python subset. Package, file and API examples may use simulated responses. Run the project locally for real integrations.

Chapters

CH 00Start HereWho the book is for and what you will have built by the end.6 min read CH 01Python by ExampleVariables, collections, functions, classes - every idea as data in, data out.18 min read CH 02Why pyproject.toml ExistsManifests, lock files and virtual environments, explained by the problem they solve.12 min read CH 03Backend FundamentalsHTTP, JSON and servers - the pipe every model call travels through.15 min read CH 04LLM ConceptsTokens, context windows, temperature and the chat message format.13 min read CH 05LM Studio, Ground UpA real OpenAI-compatible backend running on your own machine.12 min read CH 06Your First AgentMemory, tools and the think/decide/act/observe loop.14 min read CH 07Multi-Provider ArchitectureOne contract, many adapters - swap engines without touching the loop.13 min read CH 08Capstone ProjectAssembling the full modular, enterprise-shaped agent project.14 min read CH 09Memory & Context EngineeringContext rot, MemGPT/Mem0/Zep, compaction, sub-agents and the Hermes tool-calling format.20 min read CH 10RAG, GraphRAG, CAG & MCPHybrid retrieval, rerankers, agentic and corrective RAG, DAG pipelines, KV-cache corpora, MCP.16 min read CH 11Production EngineeringBackoff, idempotency, deadlines, evals, guardrails, sandboxing - why production is a different craft.12 min read

Live Python cells

A real lexer, parser and interpreter written in TypeScript executes the book's samples in your browser.

Provider adapters

Learn a shared interface for providers. The browser demonstrates that interface using simulated responses.

Inspect the tool loop

The teaching runtime chooses from local tools using deterministic rules. The chapters explain how to connect a model.

Modular services

services/python and services/agent are isolated, testable modules with no UI coupling.

Deployment practice

Study container and configuration examples, then test them with your own environment before deployment.

Print & mobile

Read on a small screen or use the print view for a quieter reference.

HELP SHAPE THE NEXT VERSION

Would you use this beyond the demo?

Tell me what you would build or learn. Your feedback will help decide whether the Python and agent sandbox should become a standalone product.

No payment or account required. I’ll use your message to understand what you need and may follow up by email. A launch is not yet confirmed.