MergeWatch vs GitHub Copilot code review

Verified April 2026. Sourced from github.com/features/copilot and docs.github.com Copilot code review. Verify current details against those pages before purchasing.

At a glance

	MergeWatch	GitHub Copilot code review
License	AGPL-3.0	Closed-source, bundled in Copilot
Self-host	Docker + Postgres, any cloud	No (GitHub.com infrastructure)
BYO LLM	Bedrock, Anthropic, LiteLLM, Ollama	No — uses Copilot’s internal selection
Agent architecture	6 review + 2 utility agents, parallel	Not publicly documented
Trigger	PR webhook, `@mergewatch`, Checks re-run	Manual or auto-assigned reviewer, IDE, `gh` CLI
Pricing	Usage-based via Stripe balance	Included in Copilot Pro/Business/Enterprise; premium requests for unlicensed org members
Data retention	Diff in-memory; 90-day metadata TTL	Business/Enterprise: 28-day prompt retention, 2-year engagement data

What it is. A pull-request review feature built into GitHub Copilot that “provides contextual explanations and code suggestions to help developers fix vulnerabilities in code.” (docs.github.com)
Trigger model. Manual by default — open the Reviewers menu on a PR and select Copilot; reviews typically complete “in less than 30 seconds.” Can be configured to auto-review all PRs. Works across GitHub web, VS Code, Visual Studio 17.14+, GitHub Mobile, Xcode, JetBrains IDEs, and gh CLI. (docs.github.com)
Where it runs. SaaS only (github.com infrastructure). (github.com/features/copilot)
LLM flexibility. The exact model powering Copilot code review is not specified on the public pages. Copilot-as-platform uses models from OpenAI, Anthropic, Google, and Microsoft, but the specific one for review is not publicly documented.
Open source. Closed-source.
Agent architecture. Not publicly documented.
Pricing. Included in Copilot Pro, Business, and Enterprise. For non-licensed users accessing it through an org, usage bills as “premium requests.” (github.com/features/copilot)
Data handling. For Business/Enterprise tiers on github.com, prompts and suggestions are retained for 28 days; user engagement data for two years.

You control the infrastructure. Copilot review runs on GitHub.com; your code flows through Microsoft. MergeWatch self-hosted keeps the PR diff and all review metadata inside your own VPC / cluster.
You choose the model. Copilot’s review model is a black box. MergeWatch lets you route to Bedrock Claude Sonnet, Anthropic direct, any OpenAI / Azure / Gemini / Mistral model via LiteLLM, or a local Ollama model. Swap with one env var.
You can audit the review logic. AGPL-3.0: every prompt, every orchestration rule is public. Copilot’s review behavior is not documented.
Documented multi-agent pipeline. Six specialized review agents (security, bug, style, error handling, test coverage, comment accuracy) plus two utilities (summary, diagram) run in parallel. Copilot’s internal composition is not disclosed.
Merge-readiness score. MergeWatch surfaces a 1–5 composite. Copilot review produces inline suggestions without a composite gate.
No per-seat Copilot license required. MergeWatch is a direct GitHub App install; reviews cost only your LLM provider + an optional managed-SaaS margin.

You already pay for Copilot Business or Enterprise org-wide and want zero additional setup — Copilot review is already on.
Your team relies on deep IDE integration across VS Code, Visual Studio, JetBrains, and Xcode, and wants review in the same surface as inline completion.
You want reviews to complete “in less than 30 seconds” on small PRs without managing LLM provider accounts.
You have strict GitHub-sole-vendor policies and can’t introduce new third-party apps with contents: read.