For working software engineers who want dependable, everyday workflows with AI coding assistants and agents across understanding, changing, testing, reviewing and shipping code.
The curriculum is being prepared and reviewed. Enrollment is not open yet.
About this course
This course is about how you work with AI assistants rather than which tool to buy (ven-190 compares the tools). It treats GitHub Copilot, Claude Code, Cursor, OpenAI Codex and Gemini CLI as interchangeable surfaces for a small set of durable workflows. You will learn to pick the right interaction mode for a task, write context files (AGENTS.md, CLAUDE.md, Copilot custom instructions) that steer agents toward your conventions, and connect issue trackers, docs and databases through the Model Context Protocol.
The core modules follow the lifecycle of a change: onboarding to an unfamiliar codebase, plan-then-implement with small diffs and checkpoints, multi-file refactors and migrations, test-first prompting with the agent running the suite, debugging from traces, logs and bisects, AI-assisted pull request review, and agents in GitHub Actions or other CI in headless mode. The security module covers secrets and permission modes for local agents, hallucinated packages and slopsquatting, and the insecure patterns models repeat. The last module asks you to measure your own productivity honestly and manage cost, rate limits and model choice. Tool specifics are as of October 2026; leading teams through this change is covered separately in wrk-203. Completion and a passing exam score earn a Wirewalk Institute of AI certificate of completion and assessment; it is not a vendor certification or endorsement.
What you will be able to do
Select an interaction mode (completion, chat, agent, background agent) appropriate to a coding task's size and risk
Write repository context files and connect MCP servers so an agent follows project conventions and has the right context
Drive a multi-file change with a plan, small reviewable diffs and passing tests
Use AI to generate meaningful tests and to debug from stack traces, logs and bisects
Configure and critically evaluate AI pull request review and agents running in CI
Apply sandboxing, secret handling and dependency checks that prevent common agent-introduced security failures
Measure the effect of AI tools on your own throughput and quality rather than relying on impressions
Syllabus
6 modules · 17 lessons · about 12 hours
1
Module 1: Working Model and Context
Interaction modes, context files and MCP connections that make agents useful.3 lessons
Completion, Chat, Agent Mode and Background Agents: Choosing the Interaction
Context Files That Steer Agents: AGENTS.md, CLAUDE.md and Copilot Instructions
MCP Servers for Issue Trackers, Docs and Databases
2
Module 2: Understanding and Changing Code
From reading an unfamiliar codebase to large refactors.3 lessons
Onboarding to an Unfamiliar Codebase with an Agent
Plan-Then-Implement: Specs, Small Diffs and Checkpoints
Multi-File Refactors and Framework Migrations
3
Module 3: Tests and Debugging
Making the agent prove its work and find real bugs.3 lessons
Test-First Prompting and Making the Agent Run the Suite
Debugging from Stack Traces, Logs and Git Bisect
Edge Cases and Property-Based Tests the Model Would Not Write Unprompted
4
Module 4: Review and Delivery
AI in pull requests, documentation and CI.3 lessons
AI Pull Request Review: Copilot Code Review and Agent Reviewers
Commit Messages, Changelogs and Docs That Match the Code
Agents in CI: GitHub Actions and Headless Modes
5
Module 5: Security and Quality Risks
The failure modes specific to AI-written code and local agents.3 lessons
Secrets, Sandboxes and Permission Modes for Local Agents
Hallucinated Packages and Slopsquatting: Dependency Hygiene
Insecure Patterns Models Repeat and Catching Them with SAST
6
Module 6: Measuring Your Own Productivity
Honest measurement, cost and model choice.2 lessons
Where Agents Slow You Down: Evidence and Honest Self-Measurement
Cost, Rate Limits and Model Selection for Daily Work
Exam and certificate
The exam unlocks once you have completed every lesson. It draws 20 questions from a larger bank, spread across the modules, with the answer options shuffled for each attempt. You have 45 minutes and need 80% to pass. After a failed attempt you can try again after 24 hours, up to 3 attempts in total. Feedback is given per module, never as the answer key.
Passing earns a certificate of completion and assessment from the Wirewalk Institute of AI with a unique ID that anyone can check at institute.wirewalk.com/verify. It is not an accredited qualification or a vendor certification. How certification works.
The Wirewalk Institute of AI is independent. This course teaches publicly documented products and is not affiliated with or endorsed by GitHub, Microsoft, Anthropic, OpenAI, Cursor, or Google. Product and company names are trademarks of their respective owners.
Practical AI for finance and accounting: reviewing models in Excel with Copilot and Claude, reconciliation and close support, invoice extraction, FP&A commentary, and the SOX and audit-evidence discipline that keeps AI-assisted numbers reliable.
Using AI in legal work without ethical or evidentiary missteps: ABA Formal Opinion 512 and the Model Rules, grounded research in Westlaw and Lexis, contract review, eDiscovery, court AI-disclosure orders and building a practice-group AI policy.
AI across the employee lifecycle with the legal guardrails built in: drafting and screening in recruiting, NYC Local Law 144, Illinois HB 3773 and the EU AI Act's employment rules, handbook assistants, performance writing, and employee privacy.