Matt Kozak

AI-Assisted Development Kit

A practical workflow and editable templates for using a coding assistant to implement features, investigate bugs, refactor code, and verify changes in an existing application.

Use AI on your codebase.
Verify the change.

A practical process for developers and technical leads using a coding assistant to implement features, investigate bugs, and refactor an existing application.

Launch price: €79

Explore the kit

Three editable documents · Getting-started guide · Checkout coming soon

01 / The problem

The diff looks plausible.
Does the behavior hold up?

A coding assistant can produce a convincing explanation before the change has been verified.

How do I give the assistant useful repository context?
How do I keep a change within scope?
How do I test the actual bug or requirement?
How do I distinguish a debugging hypothesis from an observed cause?
What does a reviewer need to assess the final change?

The kit gives you a repeatable way to frame the task, work in reviewable steps, and check the final version against the requested behavior.

02 / The outcome

Bring a real repository task.

A feature to implement. A bug to reproduce. A refactor to verify.

write a brief with scope, contracts, and acceptance criteria
trace the relevant flow before changing code
implement focused changes using the project's conventions
verify the reproduction and meaningful boundary cases
inspect the final diff and actual check results
prepare a handoff with evidence and remaining limitations

03 / What's inside

Three documents.
One development workflow.

01 / Editable brief

Context and Task Brief

Define the requested behavior, repository context, constraints, output, and verification before implementation.

Compact and full templates · Feature, debugging, and refactoring examples · Focused revision requests

02 / Development workflow

AI-Assisted Development Workflow

Inspect the relevant code, plan a focused change, implement, verify, review the diff, and hand it over.

Reusable prompts · A complete CSV export example · A failure that a superficial check would miss

03 / Review and verification

Code Review and Verify Checklist

Thirty checks covering general readiness, code changes, debugging explanations, and the review handoff. Choose the checks that matter for the task and record the evidence.

Pass / Fail / Not verified / N/A · Three worked findings and correction requests · Editable review record

Includes a getting-started guide and three editable DOCX files. Examples are illustrative; they do not report checks performed in your repository.

04 / From the kit

A passing check can miss the requirement.

Illustrative CSV export example

The page shows 25 orders. The active filters match 75 orders across three pages. The requirement is to export all matching orders in the current sort order.

A file with 25 well-formatted rows can pass a formatting check and still fail the task. The workflow compares the exported IDs with the expected matching set, including order and excluded records.

Define the behavior in the brief. Follow it through implementation. Verify the final result against the requirement.

05 / Who it's for

Developers and technical leads working on existing software.

You have a repository, a specific task, and a coding assistant that can access the relevant context. You want a practical way to guide the work and assess the result.

The kit focuses on the working method. You supply the application knowledge, tool access, and project-specific checks.

06 / Why I made it

Keep the reasoning close to the code.

Matt Kozak
Evidence → Decision → Action

More about my background →

I'm a working CTO with a background in backend systems, cloud infrastructure, connected devices, energy software, and technical leadership.

The kit applies the same emphasis on evidence that I bring to technical review: define the question, inspect the relevant system, and make the next action explicit.

07 / Your next change

Use the process on your next code change.