Product
Platform Agentic Code Scanner Unified Runner Tomosu Agents Indexes Solutions
Get Started
Scan your repo VS Code plugin Tomosu MCP GitHub App CodeRabbit FAQ
Resources
Case Studies Governance Impact TCO Calculator Download / Free Trial
Pricing
Plans and pricing
Company
About Blog News Contact Request Demo → Try Free Log in GitHub
Tomosu AI, VS Code · GitAgent

Agentic Code Scan

Ask Tomosu AI for suggestions right where you are already working, in VS Code, then carry the result into GitAgent and open a pull request without leaving your browser. No separate scan to configure, no CI step to wire up: the extension you already installed is the whole trigger.

Setup timeNone, if the plugin is installed
TriggerOne VS Code command
AuthTomosu account · GitHub (for the PR)
OutputA pull request on GitHub

01Overview

This is the shortest path from "I am looking at some code" to "there is a pull request fixing it." You never leave your editor to start it, and you never leave the browser to finish it. Two surfaces are involved:

Classification happens on our side, not yours. When you run a CodeScan, Tomosu's backend analyses the repository to determine whether it's agentic or not, then scans it accordingly. There's nothing to configure or choose, the extension works the same way either way.
Don't have the extension yet? Install it first from the VS Code plugin guide — everything below assumes it's already running.

02Run a CodeScan in VS Code

With the extension signed in, you're already looking at the Tomosu AI Assistant panel. Running a scan is one button.

1
Click CodeScan

Click the CodeScan button at the bottom of the panel. Tomosu detects the files in your workspace and runs a reliability analysis against them, read-only, it never modifies your files. A typical scan takes under three minutes.

The Tomosu AI Assistant panel after sign-in, showing the CodeScan button alongside VisionBoard, GitAgent, and the AI chat input.
Fig. 1: The Tomosu AI Assistant panel, ready to run CodeScan
2
Read the results in the panel

Results post back in the same panel: an overall PRI score out of 100, a sub-index breakdown, and recommendations prioritised by production impact, each one explaining what the issue is and why it matters before it ever reaches production.

Tomosu AI CodeScan results panel in VS Code showing a PRI score, sub-index breakdown, and prioritised recommendations.
Fig. 2: CodeScan results: PRI score, sub-index breakdown, and recommendations
This also works from the command line or from Claude. If your workflow runs through Tomosu MCP or Unified Runner instead of the VS Code extension, the destination is still the same: results land on your account, and the rest of this guide picks up from there.

03Open GitAgent

A suggestion in VS Code isn't a pull request yet, it's a recommendation against a file. Turning it into a pull request happens in GitAgent, Tomosu's git assistant.

3
Go to GitAgent and log in

Open gitagent.tomosu.ai and log in with your Tomosu account email and password.

GitAgent login screen with email and password fields, a Login button, and a Sign in with Google option.
Fig. 3: Logging in to GitAgent with your Tomosu account

04Connect Your Git Provider

The first time you use GitAgent, it needs permission to see your repositories before it can show you anything else. If you've already connected a provider on an earlier visit, GitAgent skips straight past this and you land on the repository picker below.

4
Authorise GitHub, Bitbucket, or GitLab

After logging in, connect your git provider via OAuth. GitHub is the default, but Bitbucket and GitLab work the same way. Authorise the connection on the provider's own page, and you're returned to GitAgent with your repositories available.

GitAgent Sign in to Get Started screen with Sign in with GitHub, Sign in with Bitbucket, and Sign in with GitLab buttons.
Fig. 4: Connecting your git provider, first time only

05Review and Create the Pull Request

With GitHub connected, GitAgent can pull in the fix and let you decide what happens to it. Nothing is pushed to your repository until you say so.

5
Pick the repository and branch, then fetch the fix

Select the Repository and Branch the suggestion applies to, then click Fetch Recommended change from Tomosu AI Agent App. This is what actually pulls your VS Code-originated suggestions into GitAgent — nothing shows up until you click it.

GitAgent repository and branch selectors with the Fetch Recommended change from Tomosu AI Agent App button.
Fig. 5: Repository, branch, and the Fetch button that pulls in your suggestions
6
Wait while GitAgent scans the repository

GitAgent matches the fetched suggestions against your repository's current state. A progress bar and status line show it working; this takes a few seconds to a couple of minutes depending on repository size.

GitAgent scanning the repository for suggestions, with a progress bar and status message.
Fig. 6: Scanning the repository for suggestions
7
Choose “Create Pull Request” at Deploy & Merge Options

Once the scan finishes, GitAgent asks how you want to ship the fix: open a pull request for manual review and merge, or auto-merge directly when its safety checks pass. Pick Create Pull Request to keep a human review in the loop.

GitAgent Deploy and Merge Options step, with Create Pull Request and Auto-Merge as the two deployment methods.
Fig. 7: Choosing Create Pull Request over Auto-Merge
8
Review each fix and accept the ones you want

GitAgent steps through every fix one at a time, showing the original code next to Tomosu's proposed change as a diff. Accept individual fixes, or click Accept All Pending. As you accept fixes, the panel on the right fills in the pull request itself: branch name, commit message, PR title and description, all generated from what you accepted.

GitAgent fix review screen showing a before and after code diff on the left, and the auto-generated pull request form on the right with the count of accepted fixes.
Fig. 8: Reviewing a fix, with the pull request form filling in as you accept
9
Create Pull Request with N Fix(es)

Check the PR title and description GitAgent generated, edit them if you want, then click Create Pull Request with N Fix(es). This is the moment the pull request is actually opened on GitHub, using your connected GitHub account.

GitAgent pull request description field and the Create Pull Request with 2 Fixes button.
Fig. 9: The generated PR description, and the button that opens it

GitAgent confirms once the pull request is live, with a direct link to view it on GitHub.

GitAgent success screen confirming the pull request was created, with a View on GitHub link.
Fig. 10: Done — the pull request is open on GitHub
This step is always a click, never automatic. Even when a scan arrives in GitAgent pre-filled with a repo and branch, opening the actual pull request still waits for you to review the fixes and press Create Pull Request yourself.
Request Demo →