Move your team from Amazon Q Developer to GitHub Copilot.
Each person gets a short interview, then a personalized step-by-step guide showing exactly where their saved work, prompts, and projects land in GitHub Copilot — and the few things that need a manual step, walked through. You send one invite; nobody becomes the help desk, and nothing in Amazon Q Developer is changed or cancelled. Want to see where you’re paying for both first? Run the free audit.
Both products support autonomous multi-step coding agents in the IDE that plan tasks, edit files across the workspace, run terminal commands, and present diffs for review. Amazon Q's agentic mode is on by default in the IDE chat panel (toggled with the </> icon); Copilot Agent mode is selected from the chat mode dropdown (Ask / Edit / Agent). To migrate workflows: describe your task in the same natural language in the new tool's agent prompt — no configuration files are exported. In Copilot, enable MCP servers beforehand if the task needs external tool access (same applies to Amazon Q). Accept/reject diffs in Copilot by pressing Tab (accept) or Alt+Del (reject) per file; Amazon Q uses its own diff review UI. Amazon Q is entering end-of-support (new signups blocked 2026-05-15; IDE access continues until 2027-04-30 for existing users).
- Warning: Amazon Q agentic requests are quota-metered: 50 per month on Free, shared across all agentic tasks (chat, test generation, doc generation, code review); Copilot Agent usage is metered in GitHub AI Credits by token consumption — heavy agent sessions can exhaust credits quickly on paid plans.
- Warning: Amazon Q new Free-tier signups and new Pro subscriptions were blocked on 2026-05-15; existing users retain access until 2027-04-30. Teams planning a long-term agentic-coding strategy should factor in this sunset when choosing which product to migrate to.
- Warning: Copilot Agent mode in VS Code has access to the full terminal and MCP tools; Amazon Q can run shell commands it deems low-risk autonomously — both carry risk of unintended side effects that differ in scope.
Both tools provide AI-assisted code review with inline findings and fix suggestions. Amazon Q's review runs in the IDE against recent changes (git diff, up to 200 KB) or the full project (up to 500 MB); it combines generative AI with its Amazon Q Detector Library for SAST, secrets scanning, IaC misconfiguration, SCA, and code quality. Copilot code review runs on Pull Requests on github.com and on uncommitted changes in the IDE, leaving inline comments. To migrate from Amazon Q to Copilot review: shift your review trigger to PRs on github.com and enable Copilot code review in repository settings; for IDE-side review use the Copilot 'Review and Comment' action in the IDE. To migrate from Copilot to Amazon Q: invoke review from the IDE chat panel, specifying auto-review (recent git diff) or full-project scope; apply fix suggestions directly in the editor.
- Warning: Amazon Q's code review includes a dedicated security scanner (SAST, secrets, IaC misconfiguration, SCA) backed by Amazon CodeGuru detectors — Copilot code review does not offer an equivalent depth of security-specific scanning; security-focused teams should verify coverage before switching.
- Warning: Amazon Q review is unavailable in Visual Studio; Copilot IDE code review is available in VS Code, Visual Studio, JetBrains, and Xcode — teams on Visual Studio who depended on Amazon Q review have no IDE-side equivalent issue.
- Warning: Copilot code review is in public preview and subject to change; it additionally consumes GitHub Actions minutes when triggered on PRs, adding cost beyond AI Credits.
- Warning: Amazon Q review is consumed from the same 50 agentic requests/month quota on Free as all other agentic features; heavy review users on Free tier will run out quickly.
Amazon Q's customizations fine-tune the inline suggestion model against a private codebase uploaded to S3 or connected via AWS CodeConnections; GitHub Copilot's repository custom instructions inject Markdown-based guidance into Chat, Edit, and Agent requests without retraining the model. These features address overlapping goals (aligning AI output to internal code style and APIs) but via different mechanisms. Moving from Amazon Q customizations to Copilot: document your internal library patterns, coding conventions, and API usage examples in .github/copilot-instructions.md (repository-wide) and .github/instructions/*.instructions.md (path-specific); add example code snippets to the instruction files to replicate the signal that the Q customization encoded from your private repo. Moving from Copilot instructions to Amazon Q customizations: export internal code to S3 or connect via AWS CodeConnections, create a customization in the Amazon Q console, activate it in the IDE. The Amazon Q approach requires Pro subscription and admin access to AWS.
- Warning: Amazon Q customizations require a Pro subscription and admin setup in the AWS console; Copilot repository instructions are available on all plans and require no admin infrastructure — the operational overhead is dramatically different.
- Warning: Amazon Q customizations retrain the model on private code (up to 8 customizations, 2 active simultaneously); Copilot instructions are prompt-injected text — they do not adapt the underlying model weights, so suggestions for deeply internal APIs will be less accurate without explicit examples in the instructions file.
- Warning: Copilot code review reads only the first 4,000 characters of any instruction file — long or complex instruction sets will be silently truncated for code-review requests.
Both products embed a persistent chat panel inside VS Code, JetBrains, Eclipse, and Visual Studio. To move from Amazon Q chat to Copilot Chat: install the Copilot extension, sign in, and open the Copilot Chat panel (Ctrl+Alt+I in VS Code). Re-establish any context habits — Copilot Chat uses @workspace, @file, #file, and #editor references instead of Amazon Q's @file/@folder/@workspace context keywords; slash commands differ (/fix, /explain, /tests in Copilot vs. /clear, /compact, /help in Amazon Q). Neither product exports chat history, so prior conversation transcripts are not portable. To move from Copilot to Amazon Q: install Amazon Q for VS Code/JetBrains/Eclipse/Visual Studio, sign in with AWS Builder ID, and use the Amazon Q chat panel; switch context annotations from #file to @file syntax.
- Warning: Amazon Q chat is available on the Free tier with no message cap documented; Copilot Chat Free is capped at 50 messages per month — heavy chat users switching to Copilot Free will run out quickly.
- Warning: Amazon Q inline chat (pop-up at cursor) is not available in Visual Studio; Copilot inline chat IS available in Visual Studio — this matters for teams that rely on that surface.
- Warning: Copilot Chat is available on github.com web UI and GitHub Mobile, giving it chat surfaces that Amazon Q has no equivalent for.
- Warning: Amazon Q's default chat model (Claude Sonnet 4, 200k context) is fixed to two Claude Sonnet variants; Copilot Chat lets users choose from 15+ models across OpenAI, Anthropic, Google, and Microsoft.
Both tools deliver real-time ghost-text completions in VS Code, JetBrains IDEs, Eclipse, and Visual Studio; Tab accepts a suggestion in both. To switch from Amazon Q to Copilot: install the GitHub Copilot extension in your IDE, sign in with a GitHub account, and disable or uninstall the Amazon Q extension to avoid conflicts. To switch the other direction: install the Amazon Q extension, sign in with an AWS Builder ID (Free) or AWS account (Pro), and disable the Copilot extension. No completion history or accepted-suggestion data exports exist in either product, so there is no migration artifact — just re-authentication. Language coverage is comparable (20+ shared languages); IaC is well-covered by both (Terraform/HCL, CloudFormation/YAML). Amazon Q Free has no stated monthly cap on inline completions; GitHub Copilot Free caps at 2,000 completions/month.
- Warning: GitHub Copilot Free is capped at 2,000 completions per month; Amazon Q Developer Free has no documented completion cap — teams on Free tiers moving to Copilot may hit the monthly limit.
- Warning: Amazon Q offers inline completions in AWS-specific coding environments (Cloud9, Lambda console, SageMaker Studio, JupyterLab) that Copilot does not support; workloads using those surfaces have no Copilot equivalent.
- Warning: GitHub Copilot allows per-session model selection for completions (VS Code, Visual Studio, JetBrains); Amazon Q's inline-suggestion model is not user-selectable.
Both Amazon Q Developer and GitHub Copilot support MCP (Model Context Protocol) servers to extend the agent with custom tools and data sources. Server configuration is the migration artifact. Amazon Q stores MCP config in ~/.aws/amazonq/default.json (global) or .amazonq/default.json (workspace); it also supports legacy ~/.aws/amazonq/mcp.json. Copilot stores MCP config in per-IDE settings files or repo-level config files depending on the IDE. To move MCP servers from Amazon Q to Copilot: copy the server definitions (command, args, env, transport) from the Amazon Q config into the equivalent Copilot IDE settings (e.g. VS Code settings.json under 'github.copilot.mcpServers'); verify each tool shows up in the Copilot chat tools panel. To move in the other direction: copy server definitions into ~/.aws/amazonq/default.json under the 'mcpServers' key; use the Amazon Q IDE GUI (tools icon in the chat panel) to verify each server is connected and set per-tool permissions (Ask / Always allow / Deny). Both support STDIO and HTTP transport.
- Warning: Amazon Q's HTTP MCP servers that require auth trigger an in-browser authorization flow; Copilot's per-IDE MCP setup has different auth flows depending on IDE — test auth before declaring migration complete.
- Warning: GitHub Copilot Extensions (the older App-based extensibility) were sunset on 2025-11-10 and replaced by MCP; if your team used Copilot Extensions rather than MCP servers, those configurations are not portable to Amazon Q at all.
- Warning: Amazon Q's tool-permission model is granular per tool (Ask / Always allow / Deny) inside the IDE GUI; Copilot does not expose the same per-tool permission UI — teams with strict security requirements on tool approvals should verify Copilot's approval UX meets their needs.