GitHub Copilot
by GitHub / Microsoft 4.5 / 5The lowest-friction AI completion in your existing editor.
- Best for
- Inline completion while writing YAML, Bash, Python, Terraform — plus agent mode for scoped multi-file changes
- Pricing
- Free tier (limited); Pro $10/mo; Business $19/seat/mo; Enterprise $39/seat/mo
- Vendor
- GitHub / Microsoft
Pros
- Inline completion just works — no context switching
- Wide editor support (VS Code, JetBrains, Neovim, Visual Studio)
- Agent mode and the coding agent now handle scoped multi-file changes and PR-style tasks
- Copilot CLI brings the same assistant to the terminal for shell/Git/ops one-liners
- Model choice (Claude, GPT, Gemini) — pick the model per task
- Often bundled or discounted via existing GitHub Enterprise seats
Cons
- Repo awareness improved with agent mode but still trails Cursor for large refactors
- Quality on raw infrastructure prose (postmortems, runbooks) is lower than chat tools
- Free tier is rate-limited; serious use needs at least Pro
If you already pay for GitHub, Copilot is the lowest-friction AI you can adopt. It lives in your editor, completes as you type, and gets out of the way when you don’t want it.
Where it shines for DevOps
- YAML completion. Helm values, Kubernetes manifests, GitLab CI — Copilot’s pattern matching is genuinely useful for the boilerplate-heavy parts.
- Bash one-liners and small Python scripts in the middle of larger work.
- Inline “explain this” via chat. Highlight a regex or PromQL, ask “what does this do?” — works well.
- Tests. Decent at scaffolding
batstests for Bash andpytestfor Python. - Agent mode & Copilot CLI. The newer agent mode handles scoped multi-file changes, and the CLI brings the same assistant to the terminal for Git, shell, and ops one-liners — review the diff/command, same as any AI change.
Where to be careful
- For full-file generation across an entire IaC module, Cursor’s multi-file context is stronger.
- For prose work (runbooks, postmortems, decision docs), reach for Claude or ChatGPT instead — Copilot is tuned for code, not narrative.
- The model occasionally completes with patterns from old training data that don’t match your codebase conventions. Set up
.github/copilot-instructions.mdif you care about style consistency.
How to get the most out of it
- Lean on the chat sidebar more than most engineers realize.
- Use
/explain,/fix,/testsslash commands inside the chat. - For longer reasoning, paste into a separate Claude/ChatGPT window.
Privacy
Copilot Business and Enterprise tiers do not retain prompts or use them for training. If you’re working with proprietary code, those are the relevant tiers.
Pricing notes
Free tier is decent for evaluation. Pro at $10/mo is the cheapest serious AI coding tool on this list. If your employer already has GitHub Enterprise, ask whether Copilot is bundled — it often is.