MergeWatch is an open-source GitHub App that reviews pull requests using a multi-agent AI pipeline. Self-host it anywhere with Docker and Postgres, or let MergeWatch run it for you as a managed SaaS. You choose the LLM. There is no per-seat pricing. The project is licensed under AGPL v3.Documentation Index
Fetch the complete documentation index at: https://docs.mergewatch.ai/llms.txt
Use this file to discover all available pages before exploring further.
How it compares
| MergeWatch | Greptile | CodeRabbit | |
|---|---|---|---|
| Source | Open source (AGPL v3) | Closed source | Closed source |
| Hosting | Self-hosted (Docker, any cloud) or SaaS | Enterprise self-host in AWS | Enterprise self-host only (500-seat minimum) |
| Pricing | No per-seat pricing — pay your LLM provider | 1 per review over 50 | 48/seat/month Pro Plus (annual) |
| 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 April 2026. Visit each vendor’s website for current information. For a full breakdown across eight tools, see Comparisons.
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.
Full comparisons
Side-by-side breakdown of MergeWatch against CodeRabbit, Greptile, GitHub Copilot code review, OpenAI Codex, Claude Code, Cursor BugBot, and Qodo Merge — sourced from each vendor’s own docs.
Feature matrix
Checkmark grid across ~25 features — licensing, deployment, LLM flexibility, review pipeline, triggers, and pricing.
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.