Coming soon
GitHub Copilot is not supported yet. This page describes how it will work. Today MegaLens works with Claude Code, Codex CLI, Cursor, Gemini CLI and Lovable.
MegaLens + GitHub Copilot COMING SOON
Copilot writes and reviews. When it launches, MegaLens will bring a second set of eyes.
What Copilot lacks today
Copilot is a strong coding assistant, but its review workflow has structural gaps that a second tool can fill.
One model at a time
Copilot lets you pick from a long list of models, but each review still runs on whichever one you picked. Switching models gives you a second opinion only if you ask the same question twice and compare the answers yourself — and nothing shows you where the two disagreed.
Same agent generates and reviews code
The model that wrote the code is the same one reviewing it. A model reviewing its own code is not an independent check.
Cannot "Request Changes" on PRs
Copilot can comment on pull requests, but it cannot submit formal "Request Changes" reviews. Comments are easy to skip. Structured review findings are harder to ignore.
How MegaLens will fill it
When it launches, MegaLens will run as an MCP server next to Copilot in VS Code. Same editor, different job.
MCP server alongside Copilot in VS Code
MegaLens will register as an MCP server in your VS Code workspace. Copilot will handle generation. MegaLens will handle independent review.
Models from different companies review independently
Instead of one model checking its own work, MegaLens selects up to four AI models from different companies to review your code. Different training, different blind spots, broader coverage.
A final check reviews their findings
A final check filters out weak findings. Findings are checked before they reach your editor, and the report says when that check did not run.
Consolidated, structured findings
Findings include a severity and a next step, and a file and line when you sent the code. Structured output your IDE can act on, not a wall of text.
Setup (early access, not yet supported)
GitHub Copilot is not supported yet. If you want to try it early, the setup wizard can write the config. It only finds VS Code when .vscode/mcp.json already exists in the folder you run it from.
npx megalens-mcp setup
Or paste it yourself:
.vscode/mcp.json
{
"servers": {
"megalens": {
"type": "http",
"url": "https://megalens.ai/api/mcp?ide=copilot",
"headers": {
"Authorization": "Bearer ml_tok_your_token_here"
}
}
}
}Get your token at megalens.ai/app/settings/mcp. Setup wizard source: github.com/megalens/mcp.
Honest limitations
We'd rather you know this upfront than find out after setup.
Advisory only
MegaLens surfaces findings and recommendations. It does not override Copilot's model selection or intercept its completions. You decide what to act on.
Reviews surface as suggestions in chat, not inline diff comments
Copilot's MCP integration exposes tools through the chat panel. Review findings will appear there as structured suggestions, not as inline annotations in the editor diff view.
MegaLens answers when asked
MegaLens can respond to review requests but cannot proactively inject context into Copilot's workflow.
GitHub Copilot support is coming soon.
Until then, MegaLens works with Claude Code, Codex CLI, Cursor, Gemini CLI and Lovable.