How it compares
| MergeWatch | Greptile | CodeRabbit | |
|---|---|---|---|
| Source | Open source (AGPL v3) | Closed source | Closed source |
| Hosting | Self-hosted (Docker, any cloud) or SaaS | Self-hosted available (Docker, any cloud) | Self-hosted available (Enterprise, 500-seat min, $15K+/mo) |
| Pricing | No per-seat pricing — pay your LLM provider | Base platform fee + usage-based pricing | 30/seat/month (annual vs monthly) |
| Review approach | Multi-agent parallel pipeline | Full-codebase graph + multi-hop analysis | AI review + 40+ linter integrations |
| Model control | You choose the LLM provider and model | Bring-your-own-LLM (self-hosted) | Vendor-controlled (SaaS); configurable on self-hosted |
| Minimum commitment | None | None stated | 500 seats for self-hosting |
Competitor pricing and features were last verified in March 2025. Visit each vendor’s website for current information.
Key differentiators
No per-seat pricing
Install on your entire org. Costs scale with LLM usage, not headcount. A 200-person team pays the same platform cost as a 5-person team.
Your infrastructure, your data
Self-host MergeWatch with Docker on any cloud or bare metal. Your code stays on your servers. You see every log. MergeWatch (the company) never sees your code.
AGPL v3 — genuinely open source
Read the source, audit it, fork it, contribute back. No open-core bait-and-switch. The AGPL v3 license means improvements must be shared, which keeps the project honest.
Multi-agent parallel review pipeline
Eight specialized agents — security, bugs, style, error handling, test coverage, comment accuracy, summary, and diagram — run in parallel. An orchestrator agent deduplicates overlapping findings and ranks them by severity before posting to GitHub.
The review pipeline at a glance
Each agent is a separate LLM invocation running in parallel via
Promise.all(). The orchestrator runs after all agents complete.Deployment models
| Model | Who runs it? | Data stays where? | Setup |
|---|---|---|---|
| Self-Hosted | You | Your infrastructure | docker-compose up |
| Managed SaaS | MergeWatch | MergeWatch AWS + your GitHub | GitHub App install |
Infrastructure
Self-Hosted Stack
MergeWatch self-hosted runs as three Docker containers from a singledocker-compose.yml:
- Express server — port 3000, receives GitHub webhooks and runs the review pipeline
- Next.js dashboard — port 3001, browser UI for reviews, repos, and settings
- PostgreSQL 16 — review state, repo config, installation data
- Your LLM provider — Anthropic (default), LiteLLM (100+ providers), Amazon Bedrock, or Ollama
SaaS Stack
The managed SaaS runs on a serverless stack in MergeWatch’s AWS account:- AWS Lambda — WebhookHandler (512 MB / 30 s) and ReviewAgent (1024 MB / 300 s)
- Amazon DynamoDB — review state, repo config, user data
- API Gateway — GitHub webhook receiver
- Amazon Bedrock — Claude Sonnet (
us.anthropic.claude-sonnet-4-20250514-v1:0)
.mergewatch.yml file at the root of each repository. The dashboard provides a UI for monitoring reviews, managing repos, and adjusting settings.
How is this different from CodeRabbit?
How is this different from CodeRabbit?
Both tools review PRs with AI. The differences are structural:Where your code goes. CodeRabbit’s SaaS sends your code to their servers. Enterprise self-hosting exists but requires a 500-seat minimum at $15K+/month. MergeWatch self-hosts via a single
docker-compose up with no seat minimum — code never leaves your infrastructure.How you pay. CodeRabbit charges 30/seat/month depending on billing term. MergeWatch has no per-seat fee. You pay your own LLM provider directly, which scales with usage, not team size.Source availability. CodeRabbit is closed source. MergeWatch is AGPL v3 — you can read every line of the review logic, audit exactly what runs on your code, and contribute improvements.Model choice. On CodeRabbit SaaS, the vendor controls model selection. MergeWatch lets you choose any LLM provider — Anthropic, LiteLLM (100+ providers), Amazon Bedrock, or Ollama — and swap models at any time.Quickstart
Install MergeWatch and get your first review in under 10 minutes.
How it works
Detailed walkthrough of the review pipeline, agent architecture, and orchestration.