AI for software developers
Use AI coding tools to write, test and review code faster, while staying responsible for what ships.
From a first session with an AI coding assistant to agents, tool connections and code review, ending with a change you shipped and checked yourself.
Foundations
What you need to know before you build.
- Infrato coursePrompting Fundamentals45 min
Start with the habits that carry over to coding assistants: outcome, context, format, review.
- Infrato courseUse AI Safely at Work: Privacy and Data55 min
Know what you paste into an assistant: sort your data, strip what you can and find out where it goes.
- Getting started with GitHub Copilot
GitHub · CourseFreeGitHub’s own first course on its coding assistant.
- LLM Course
Hugging Face · CoursePrice not listedHow large language models work, for developers who want more than the surface.
Build
Make the thing.
- Infrato courseAI-Assisted Coding You Can Trust60 min
Work in small reviewable steps, review assistant code for invented packages, and ship a verified change.
- Infrato courseGive AI What It Needs: Context Design55 min
Give an assistant what it needs and no more: choose the context, place it, and keep a long session fresh.
- Build applications with Copilot agent modeGitHub · CourseFree
Building an application with the assistant in agent mode.
- Test and debug with GitHub CopilotGitHub · CourseFree
Using the assistant for tests and debugging, where you still own the result.
Extend
Check it, harden it, go further.
- GitHub Copilot code reviewGitHub · CourseFree
Reviewing code with an assistant, and what a human still has to check.
- Infrato courseDelegate to AI Agents Without Losing Control55 min
Let an agent act without losing control: least access, approval for high-impact actions, and a record of what it did.
Tools to try
Products from each vendor’s own site, listed by name and link only. We have not tested them, they are not endorsements, and we are not affiliated with the vendors. Prices and features change, so check the vendor’s page.
GitHub
GitHub CopilotGitHub’s AI coding agent.
Anthropic
Claude CodeAn AI coding agent for the terminal and IDE.
Cursor
CursorAn AI coding agent.
Google
Gemini Code AssistGemini Code Assist for teams and businesses.
There are more by what you use them for: AI tools worth learning.
Agents and MCP servers for this role
An MCP server runs code with your accounts and can act in them. Read what each one can reach, and which credentials it asks for, before you connect it. These are picked from our directories by hand, not ranked, and are not endorsements.
MCP servers
- servers
The protocol’s own collection of reference servers.
Reaches: github. Asks for: GITHUB_PERSONAL_ACCESS_TOKEN.
- context7
Up-to-date code documentation for AI coding tools.
Reach not listed. No credentials listed.
- firecrawl-mcp-server
Web scraping and search for an assistant.
Reaches: firecrawl. Asks for: FIRECRAWL_API_KEY.
- dbhub
Connects an assistant to several database servers.
Reaches: mysql, postgres, sqlite. No credentials listed.
- git-mcp
Lets an assistant read a GitHub repository’s docs and code.
Reaches: github. No credentials listed.
Agent skills
- code-review
Reviews changes against coding standards and requirements.
- tdd
Supports test-driven development.
- diagnosing-bugs
A diagnosis loop for hard bugs and performance regressions.
- claude-api
A reference for the Claude API and SDK.
Your project
Ship one change with an AI pair
Pick a small issue in a repository you own. Use an AI coding assistant to write the change and its tests, then review every line yourself. Keep a short list of what the assistant got wrong or missed, and what you would check next time.
The order and the notes are Infrato's own picks. The resources are made by their providers, open at their own sites, and can change; prices, levels and durations are the provider's. Infrato is not affiliated with the providers listed.