Finds the bugs.
pens the fix.
Reviews for AI-powered teams who move fast (but don't break things)
527
PRs reviewed
2,143
bugs found
29s
avg review
Good morning
Developer
94%
PRs pass clean
4.2
avg issues/PR
47
7d reviews
312
30d reviews
12
repos
VS Code Extension
Review code directly in your editor before pushing
MCP Integration
Use with Claude, Cursor, Kiro, or any MCP-compatible agent
Good morning
Developer
94%
PRs pass clean
4.2
avg issues/PR
47
7d reviews
312
30d reviews
12
repos
VS Code Extension
Review code directly in your editor before pushing
MCP Integration
Use with Claude, Cursor, Kiro, or any MCP-compatible agent
Setup
Three steps. Zero config.
Install the GitHub App
One click from the marketplace. No CI/CD changes needed.
Open a pull request
CodePeel triggers automatically — no commands, no webhooks to configure.
Get your review
Security score, findings, walkthrough, and auto-fix PRs — typically under a minute.
The Payoff
It doesn't just find the bug.
It opens the fix.
Every other reviewer leaves you a comment. CodePeel writes the patch, opens a real PR, and waits for your merge.
Bug Found
auth/login.ts:24
Fix Written
AST Auto-repair
PR Opened
codepeel/fix-auth-sql
Merged to Main
Health: 41 → 96
Features
Review. Fix. Test. Ship.
Free users get the full review engine. Pro adds the fixes and tests.
Auto-Fix PRs
When CodePeel finds a fixable issue, it opens a real PR with the fix applied to your branch. Review the diff, merge, done.
Three Engines, One Review
Every PR runs through bug detection, OWASP security scanning, and architecture analysis in parallel. Findings are severity-rated.
Auto-Generated Tests
Generates framework-appropriate tests (Jest, Vitest, pytest, flutter_test, Go, Rust) and opens them as a PR. Covers happy paths and error cases.
Pre-Merge Quality Gates
Commit statuses that fail if critical findings exist. Works with your existing branch protection rules. Block bad merges automatically.
Everywhere You Code
One reviewer. Every surface.
GitHub, your editor, and your AI coding agent — all feeding into one review engine, one account, one quota.
VS Code Extension
codepeel-vscode
Review changes before you push. Inline comments, one-click fixes, and severity ratings — right in your editor.
- Review uncommitted or committed changes
- Inline comments on exact lines
- Severity ratings and health score
- Works with GitHub and local repos
MCP Server
github:codepeel/mcp-server
Let your AI coding agent review its own output. Works with Claude Code, Cline, Roo, Kiro, and any MCP client.
- AI reviews code during your coding session
- Catches bugs before you push
- Works with Claude Code, Cline, Roo, Kiro
- Simple JSON config setup
Pricing
Simple, transparent pricing for engineers & teams. Lock in early adopter pricing forever.
- 30 reviews per month
- Unlimited repos (public + private)
- AI-powered code reviews
- Security vulnerability detection
- VS Code, Claude Code & Cursor support
- 500 reviews per month
- Auto-fix pull requests
- Auto-generated unit tests
- Pre-merge quality gates
- Custom review rules
- Unlimited reviews per month
- Premium AI models (Claude 3.7 & GPT-4o)
- Priority review execution queue
- Custom prompts & architecture guides
All plans include GitHub App installation. No API key required.
Security
Your source code stays yours.
We analyze your diffs in-memory and never persist raw source code. Review results power your dashboard — your codebase never leaves GitHub.
Source code never stored
Diffs are analyzed in-memory and never persisted. We store review results (findings, scores) to power your dashboard — not your source code.
No AI training
Your code is never used to train any model. We use zero-retention LLM providers exclusively.
Scoped permissions
Read-only access for analysis. Write access is only used for auto-fix branches — never touches your main branch.
FAQ
Common questions.
Will CodePeel flood my PRs with noisy nitpicks and false positives?
No. CodePeel is engineered with a precision-over-recall philosophy. Slop detectors and static analysis filters discard stylistic fluff, while customizable profiles (chill, balanced, strict) and @codepeel ignore: rules let you permanently suppress unwanted noise.
Does CodePeel store, retain, or train on my proprietary code?
Never. We enforce a strict Zero Data Retention policy for source code. PR diffs are processed ephemerally in-memory to generate findings and immediately purged. Your code is never used to train or fine-tune public or private AI models.
Which AI models power CodePeel, and do I need to bring API keys?
CodePeel uses a proprietary multi-model hybrid review engine combining fast-pass syntax scanners with deep-context reasoning models. All compute, token budgets, and automatic failovers are 100% managed — no API keys or token configurations required.
Can I use CodePeel before opening a PR or with AI coding agents?
Yes. CodePeel includes an official VS Code extension for pre-push diff reviews, plus a standard Model Context Protocol (MCP) server that lets agents like Claude Code, Cursor, Cline, and Windsurf self-review code before submitting.
How does pricing work across public and private repositories?
All plans — including the 30-review/month Free tier — support unlimited public and private repositories. Pro ($9.99/mo) unlocks 500 reviews, automated fix PR generation (auto_fix), and unit test creation (auto_test) without expensive per-seat traps.
How is CodePeel different from GitHub Copilot or CodeRabbit?
While Copilot suggests code in the editor, CodePeel acts as the automated senior reviewer at your CI/CD gate. Unlike bots that only post text complaints, CodePeel generates ready-to-merge fix PRs, automated unit tests, and integrates directly via MCP.
Start catching bugs in 2 minutes.
Install the GitHub App. Open a PR. Get your first review.
Get Started Free