GitLab Setup Guide: Integrating Gito with Your Repository

July 12, 2026 · View on GitHub

Automate code review for all Merge Requests using AI. This guide shows how to connect Gito to a GitLab project for continuous, automated MR reviews using the gito deploy wizard.

Prerequisites

Maintainer or Owner access to your GitLab project.
An API key for your preferred language model provider (e.g., OpenAI, Google Gemini, Anthropic Claude, etc).
The gito CLI installed locally (pip install gito.bot). See the README for installation options.

1. Generate the CI files with `gito deploy`

From the root of your locally cloned repository, run:

gito deploy

Aliases: gito init, gito connect, gito ci.

The interactive wizard will:

Detect that the project is hosted on GitLab.
Ask which language model provider and model to use.
Write the CI workflow files (see below).
Optionally commit and push them to a dedicated gito-ci branch and open a Merge Request for you.
Print the exact CI/CD variables you still need to add.

It creates two files:

.gitlab/ci/gito-code-review.yml — the Gito review job.
.gitlab-ci.yml — a minimal pipeline that adds a review stage and includes the job above. If you already have a .gitlab-ci.yml, Gito merges the review stage and include into it instead of overwriting.

The generated .gitlab/ci/gito-code-review.yml looks like this:

# Gito AI Code Review for GitLab
gito-ai-review:
  stage: review
  image: python:3.13
  rules:
    - if: $CI_PIPELINE_SOURCE == "merge_request_event"
    - if: $CI_PIPELINE_SOURCE == "web" && $MR_NUMBER  # Manual trigger via "Run pipeline"
      when: manual
  variables:
    GIT_DEPTH: 0  # Full history needed for diff
    LLM_API_TYPE: openai
    LLM_API_KEY: $LLM_API_KEY
    MODEL: gpt-5.2
  script:
    - pip install gito.bot~=4.4
    - gito --verbose review --against="origin/$CI_MERGE_REQUEST_TARGET_BRANCH_NAME"
    - gito gitlab-comment --token "$GITLAB_ACCESS_TOKEN" --project-id "$CI_PROJECT_ID" --merge-request-iid "$CI_MERGE_REQUEST_IID"
    - gito -v0 render gitlab_code_quality > gitlab_code_quality_report.json
  artifacts:
    reports:
      codequality: gitlab_code_quality_report.json

Prefer manual setup? You can copy the GitLab workflow templates directly instead of running gito deploy.

2. Add CI/CD variables

Go to Settings → CI/CD → Variables and add:

Variable	Value	Notes
`LLM_API_KEY`	Your LLM provider API key	Name must match `LLM_API_KEY` in the generated job.
`GITLAB_ACCESS_TOKEN`	A Project Access Token	Scope `api`, role Reporter (or higher). Create at Settings → Access Tokens.

A dedicated GITLAB_ACCESS_TOKEN is required because the default $CI_JOB_TOKEN cannot write Merge Request comments.

For each variable:

☑ Mask variable
☐ Protect variable — leave unchecked, otherwise Merge Request pipelines won't have access.
For public repositories, enable Require approval for fork pipelines (CI/CD settings) to prevent secret leaks. See GitLab CI/CD variables docs.

3. Open a Merge Request

Once the CI files are on your default branch and the variables are set, every Merge Request triggers a review. You'll see:

An AI-generated overview comment on the MR.
A GitLab Code Quality report attached to the pipeline (see below).

To re-run a review, push a new commit or trigger the pipeline manually via Run pipeline (set the MR_NUMBER variable).

Reporting modes

Gito posts review results to the MR via the gito gitlab-comment command. Two modes are available:

Overview comment (default)

A single comment summarizing the review, with all detected issues listed in the body.

Inline comments — `--inline`

Add the --inline flag to post each issue as a separate inline comment anchored to the affected diff lines, while the main comment keeps only the review overview:

    - gito gitlab-comment --inline --token "$GITLAB_ACCESS_TOKEN" --project-id "$CI_PROJECT_ID" --merge-request-iid "$CI_MERGE_REQUEST_IID"

Inline comments are created as draft notes anchored to the MR diff and published together as a single review. Unlike the Code Quality artifact, inline comments work on all GitLab tiers. Issues that cannot be anchored to changed lines are included in the overview comment instead.

Code Quality artifacts

The generated job also renders a GitLab Code Quality report:

    - gito -v0 render gitlab_code_quality > gitlab_code_quality_report.json
  artifacts:
    reports:
      codequality: gitlab_code_quality_report.json

GitLab ingests this artifact and surfaces the detected issues directly in the Merge Request widget and the pipeline's Code Quality tab:

Note: The Code Quality MR widget is available on GitLab Premium/Ultimate. On all tiers, the report can still be downloaded as a pipeline artifact, and inline comments (--inline) remain available.

Customize Review if Needed

Create a .gito/config.toml file at your repository root to override the default configuration.
You can adjust prompts, filtering, report templates, issue criteria, and more — see the Configuration Cookbook.

Troubleshooting

No comment on the MR? Open the pipeline, check the gito-ai-review job logs for errors (missing LLM_API_KEY, insufficient GITLAB_ACCESS_TOKEN scope/role, etc).
403/permission errors when posting? Ensure GITLAB_ACCESS_TOKEN is a Project Access Token with api scope and at least the Reporter role, and that the variable is not Protected.
Empty diff / nothing reviewed? Confirm GIT_DEPTH: 0 is set so the full history is available for the diff.

Additional Resources

More usage documentation: README.md
Command Line Reference
For help or bug reports, open an issue

Enjoy fast, LLM-powered merge request reviews and safer merges! 🚀