Build a dependable Python foundation
Use isolated projects, explicit contracts, useful errors, and tests that cover both success and failure behavior.
Learning roadmap
Learn in the order real systems demand.
Start with solid Python, ship a small application, automate a real task, then add agent behavior where it earns its complexity.
Ordered learning path
Turn code into a working application
Define a stable HTTP contract, validate at the boundary, bound dependencies, and verify production behavior.
Read: Design a small Python web API with explicit boundaries →
Automate a real workflow
Make repeated work observable, idempotent, safe to resume, and strict about business completion.
Add an agent only where judgment is needed
Keep action validation, tools, deadlines, traces, and review outside the model.
Read: Build a testable Python agent loop without an API key →
Featured guide
Run and test the agent loop before the framework.
A no-key troubleshooting agent with bounded tools, a complete action trace, and automated boundary tests.
Build a testable Python agent loop without an API key
Build a small troubleshooting agent with bounded tools, validated actions, a complete trace, and automated tests using only the Python standard library.
Read the complete guide →First-party guides
Four paths, one reliability standard.
Every guide declares its sources, author, last verification date, and either a runnable command or an editorial review record. Subscribe via RSS →
Python foundations
Build a dependable Python foundation with tests
Learn a repeatable Python project loop built around isolated environments, explicit data contracts, failure cases, and standard-library tests.
Web & APIsDesign a small Python web API with explicit boundaries
Plan a Python API around validated input, stable responses, timeouts, observable failures, and a deployment contract before choosing more infrastructure.
AutomationBuild safe Python automation that can be retried
Turn a repeated task into an observable Python workflow with explicit inputs, idempotent writes, bounded retries, checkpoints, and recovery evidence.
AI agentsBuild a testable Python agent loop without an API key
Build a small troubleshooting agent with bounded tools, validated actions, a complete trace, and automated tests using only the Python standard library.