Skip to main content

Getting Started

Get your first AI code review in under 5 minutes.


Step 1: Sign Up

Go to codepeel.ai and click Get Started Free. Sign in with your GitHub account — no credit card required.

After signing in, you'll land on the Dashboard. This is your home base for all review activity.

CodePeel dashboard showing the empty state with a prompt to get your first AI review

Step 2: Connect Your Repositories

From the dashboard, click Connect in the top-right of the Repositories page (or follow the prompt on the empty dashboard).

This takes you to GitHub's app installation page where you choose which repositories CodePeel can access:

  • All repositories — CodePeel reviews every PR automatically
  • Selected repositories — pick specific repos to monitor

After installing, your connected repos appear on the Repositories page in the sidebar.

Repositories page showing connected repos with review stats and status

Free tier: Up to 3 repositories. Private repos require Pro or Max.


Step 3: Open a Pull Request

That's it — no more setup needed. CodePeel triggers automatically when:

  • A new PR is opened
  • A new commit is pushed to an existing PR

Note: Only PRs created after installation are reviewed. Push a new commit to an existing PR to trigger a review on it.

Within 30-60 seconds, CodePeel posts three things on your PR:

1. Inline Comments

Posted directly on the lines that need attention in the Files changed tab on GitHub:

CodePeel inline comment on a GitHub PR showing severity badge, explanation, and code suggestion

Each comment includes:

  • Severity badge — 🔴 Critical, 🟠 Major, 🟡 Minor
  • Structured explanation — "What's happening", "Impact", and "Recommendation" sections
  • Code suggestion — a one-click "Commit suggestion" button that applies the fix directly

Only high-value findings are posted as inline comments. Lower-severity findings appear in your dashboard to keep PRs clean.

2. Summary Walkthrough

A collapsible comment on the PR conversation tab with:

  • Plain-English summary of what the PR does
  • Finding counts by category (bugs, security, performance, best practices)
  • Review effort rating (1-5) — how complex the PR is to review
  • Health score (0-100) — aggregate code quality score
  • Sequence diagram for complex logic (when auto_sequence_diagram: true)
CodePeel walkthrough summary comment on a GitHub PR with finding counts and health score

3. Commit Status

A green ✅ or red ❌ status check appears on the PR:

  • codepeel/premerge — passes/fails based on your pre-merge check rules
  • CodePeel — shows total finding count
GitHub PR status checks showing codepeel/premerge passed with a green checkmark

Pre-merge checks are optional. Set them up to block merges when critical issues are found.


Step 4: Review in the Dashboard

Every review is also visible in the CodePeel webapp at codepeel.ai/app:

  • All PRs — list of all reviewed PRs with status and finding counts
  • PR Detail — full walkthrough, findings with code diffs, one-click fix buttons
  • Fix Queue — PRs with unresolved critical/high findings that need attention
PR detail page in the CodePeel dashboard showing summary, walkthrough, findings list, and checks sidebar

The dashboard also gives you access to:


Step 5: Interact with @codepeel

Mention @codepeel in any PR comment to interact with the bot. Full details: @codepeel Chat docs

Available commands:

CommandWhat it doesExample
(just ask)Ask any question about the code@codepeel is there a race condition here?
learn:Teach a rule for future reviews@codepeel learn: Always use Zod for validation
ignore:Suppress a false positive pattern@codepeel ignore: console.log in debug.ts is intentional
resolveDismiss a finding (won't re-flag)Reply @codepeel resolve on any inline comment
initGenerate .codepeel.yml config PR@codepeel init
reset configOverwrite existing config with defaults@codepeel reset config

Tip: Rules saved via learn: apply to all future reviews in that repository. For version-controlled rules, use expert_rules in .codepeel.yml. See Learnings docs for the full guide.


What Gets Reviewed

CodePeel runs four analysis layers in parallel on every PR:

LayerWhat it detectsSpeed
Secret scanningAPI keys, tokens, credentialsInstant
AI analysisBugs, security, performance, best practices~10-30s
SASTStatic analysis security patterns~10-20s
Architecture reviewDesign patterns, coupling, separation of concerns~15-25s

Finding categories:

CategoryExamples
BugsLogic errors, null references, off-by-one, unhandled promises
SecuritySQL injection, XSS, hardcoded secrets, path traversal
PerformanceN+1 queries, memory leaks, unnecessary re-renders
ArchitectureDesign patterns, coupling, separation of concerns

Want to customize what gets flagged? Use review profiles (chill, balanced, assertive) or enable securityOnly mode.


Pricing

PlanPriceReposReviews/month
Free$03 public50
Pro$9.99/user/moUnlimited500
Max$0.04/reviewUnlimitedNo limit

All reviews deduct from your monthly quota regardless of repo visibility.


What's Available Beyond PR Reviews

CodePeel isn't just a PR reviewer. Here's what else you can do:

FeatureDescriptionDocs
VS Code ExtensionReview code locally before pushingExtension docs →
MCP ServerUse CodePeel inside AI editors (Kiro, Claude, Cursor)MCP docs →
Auto-Fix PRsAutomatically generate fix PRs for findingsAuto-Fix docs →
Auto-Test PRsGenerate unit tests for your changesAuto-Test docs →
Pre-merge ChecksBlock merges when quality gates failPre-merge docs →
Knowledge BaseTeach CodePeel your team's conventionsLearnings docs →
.codepeel.ymlPer-repo configuration (ignore paths, rules, profiles)Config docs →

Next Steps

  1. Configure .codepeel.yml — customize review behavior, ignore paths, add expert rules
  2. Set up Pre-merge Checks — block merges on critical findings
  3. Install the VS Code Extension — review code before pushing
  4. Set up MCP — use CodePeel inside your AI coding agent
  5. Teach rules via @codepeel — make reviews smarter over time
← All docsCodePeel