Connect Tomosu MCP to CodeRabbit

01What is Tomosu MCP?

MCP (Model Context Protocol) is an open standard that lets AI tools query external data sources in real time. Tomosu MCP exposes your codebase's reliability data — scan reports, PRI scores, code suggestions, and ticket context — through a set of tools that CodeRabbit can call directly while reviewing a pull request.

Once connected, CodeRabbit automatically has access to:

Live reliability scores for the repositories it reviews
AI-generated code improvement suggestions from Tomosu
Historical scan data that informs the review context
Linked Jira, GitHub Issues, and Zendesk ticket context

No manual steps after setup. Once connected, every CodeRabbit review in your organization can query Tomosu data automatically — without anyone needing to paste scores or copy suggestions by hand.

02Connection Steps

1

Open Integrations in CodeRabbit

In the CodeRabbit sidebar, expand Review and click Integrations.

The CodeRabbit left sidebar with Review expanded, showing Triage, Repositories, Integrations, Learnings, and Organization settings. — Fig. 1 — Review › Integrations in the CodeRabbit sidebar

2

Open the MCP Servers tab and add a new server

On the Integrations page, click the MCP Servers tab. You'll see any servers already connected. Click + New MCP Server in the top-right corner.

The CodeRabbit Integrations page on the MCP Servers tab, showing a list of available integrations including Notion, Context7, Linear, GitHub Copilot, Sentry, Asana, and Monday.com, with a + New MCP Server button in the top right. — Fig. 2 — MCP Servers tab with the + New MCP Server button

3

Fill in Tomosu MCP details and connect with OAuth

In the Add MCP Server modal, enter the following:

Label: Tomosu AI
Server URL: https://mcp.tomosu.ai/mcp
Authentication: select OAuth

Click Connect with OAuth to continue.

The Add MCP Server dialog in CodeRabbit with Label set to Tomosu AI, Server URL set to https://mcp.tomosu.ai/mcp, and the OAuth radio button selected. The Connect with OAuth button is highlighted at the bottom right. — Fig. 3 — Add MCP Server modal with Tomosu AI details and OAuth selected

4

Authorize via Google

A popup opens and takes you through Google OAuth. Sign in with the Google account associated with your Tomosu account. The popup closes automatically once authorization is complete.

A dark screen showing a green checkmark circle with the text Integration Successful and Authorization complete! Closing window... — Fig. 4 — Integration Successful confirmation after OAuth completes

5

You're connected — 8 tools enabled

Back on the Integrations page, mcp:tomosu ai now appears at the top of the MCP Servers list with 8 of 8 tools enabled and the endpoint https://mcp.tomosu.ai/mcp confirmed.

The CodeRabbit Integrations page showing mcp:tomosu ai connected as Direct with 8 of 8 tools enabled. The tool chips suggest_code, save_recommendation, get_scan_report, score_pull_request, get_dashboard_analytics, get_tickets, get_zendesk_tickets, and get_support_agent_data are all visible. — Fig. 5 — Tomosu AI connected with all 8 tools enabled

03Tools CodeRabbit Gets

After connecting, CodeRabbit's AI can call any of the 8 Tomosu tools during a review. Each tool gives CodeRabbit a different type of context about your codebase:

Tool	What it provides
`suggest_code`	AI-generated code improvements for the changed files in a PR
`save_recommendation`	Persists a Tomosu recommendation to the VisionBoard for tracking
`get_scan_report`	Full reliability scan report for a repository — all issues, severities, and categories
`score_pull_request`	PRI score plus FI, GC, and CV sub-index breakdown for a specific PR
`get_dashboard_analytics`	Aggregated reliability metrics and trends across your repositories
`get_tickets`	Linked Jira and GitHub Issues context for a repository or PR
`get_zendesk_tickets`	Linked Zendesk support ticket context for a repository
`get_support_agent_data`	Support escalation signals — which issues are generating the most customer impact

04How It Enhances Reviews

With Tomosu MCP connected, CodeRabbit's reviews become significantly richer:

Reliability scores in every review

CodeRabbit can pull the live PRI score for the PR it's reviewing and surface it alongside its own analysis — giving your team a reliability verdict from Tomosu without leaving the CodeRabbit review interface.

Code suggestions grounded in scan data

When CodeRabbit suggests a fix, it can cross-reference Tomosu's suggest_code output for the same files. The result is recommendations that account for both style preferences and production reliability issues identified by Tomosu's AI.

Ticket and escalation context

Tomosu surfaces Jira, GitHub Issues, and Zendesk context through MCP. CodeRabbit can mention relevant open tickets or escalation signals in its review comments without anyone having to paste links manually.

Shared across your organization. MCP server connections in CodeRabbit are shared at the organization level — one team member connects Tomosu once, and CodeRabbit uses it for every review across all repositories in the organization.

Disconnect or re-authenticate

To disconnect, go to CodeRabbit → Review → Integrations → MCP Servers, find mcp:tomosu ai, and click Delete Integration. To re-authenticate without deleting, click Re-authenticate on the same card.

05What You'll See in Pull Requests

When CodeRabbit reviews a pull request with Tomosu MCP connected, it posts a Tomosu AI reliability comment alongside its own review. Here is a real example.

A CodeRabbit pull request comment showing Tomosu AI Production Reliability: PRI 75/100 Ready to Merge, a score table with FI 100, GC 75.4, CV 92, and a before/after code recommendation for adding retry backoff to an HTTP call. — Fig. — Tomosu AI reliability comment posted by CodeRabbit on a pull request

Merge verdict

A clear Ready to Merge or Not Ready to Merge verdict. The threshold is PRI ≥ 70.

Score table

Index	Full name	What it measures
PRI	Production Reliability Index	Overall reliability score, 0–100
FI	Fragility Index	Code stability — starts at 100, deducted per issue severity
GC	Governance Compliance	Best-practice rules followed (out of 10)
CV	Code Volatility	Issue count — fewer issues = higher score

Recommendations

Each recommendation includes the affected file, category, severity, and exact before/after code so you know precisely what to change.