Set up Git Forge

Setting up Git Forge involves these steps:

• Confirm the requirements.
• Prepare the repository CSV.
• Configure Git Forge in Glean.
• Validate the setup.
• Verify indexed content.

Requirements

Before you begin, gather the following:

• The HTTPS URL of the Git server that hosts the repositories, including https://.
• A personal access token that can read every repository you plan to index.
• The username to send with the personal access token for HTTP Basic authentication. The required value depends on the Git server.
• A repository CSV file with the required columns and one row for each repository to crawl.
• Network access from the Glean crawler to the Git server over HTTPS.

The Git server must support Git Smart HTTP. Git Forge does not use a provider-specific API to enumerate repositories.

Git Forge constructs each clone URL as the Git Server URL, followed by repositoryName and .git. Include any additional clone-base path required by your Git server in Git Server URL.

Prepare the repository CSV

Create a CSV with this header:

repositoryName,repositoryUrl,defaultBranch

Add one row for each repository. For example:

repositoryName,repositoryUrl,defaultBranch
org/repo,https://git.example.com/org/repo,main

CSV columns

Column	Required value
repositoryName	The server-relative repository path used to construct the clone URL. Include the organization or namespace, such as org/repo.
repositoryUrl	The browser URL of the repository. Git Forge uses this value to construct links to files, directories, and commits in Glean results.
defaultBranch	The branch name used in file and directory result URLs and branch metadata. This value must match the branch selected by the Git server during the initial clone. It does not select the branch to crawl.

Repository names must be unique. Do not include two rows with the same repositoryName.

warning

repositoryName, repositoryUrl, and defaultBranch serve different purposes. Git Forge combines Git Server URL with repositoryName to access the repository. It uses repositoryUrl and defaultBranch to build result links. Check all three values against your Git server before uploading the file.

Configure Git Forge in Glean

1. In the Glean Admin console, go to Data sources and add Git Forge.
2. In Git Server URL, enter the HTTPS clone-base URL, including https://.
3. In Personal Access Token, enter the token that can read the configured repositories.
4. In Git Username, enter the username required by the Git server for personal-access-token authentication.
5. Upload the repository CSV.
6. Save the connector.

Git Forge stores Personal Access Token and Git Username as connector secrets and uses them when it accesses repositories. Do not include credentials in the CSV or repository URLs.

Validate the setup

When you save the connector, Glean runs these validation checks:

1. The Git Username secret exists.
2. The Personal Access Token secret exists.
3. The repository CSV exists, is readable, has the exact header, contains at least one row, and has no duplicate repositoryName values.
4. The configured credentials can access the first five repositories in the CSV, or all repositories if the CSV contains fewer than five rows. This check constructs each clone URL from Git Server URL and repositoryName.

Setup validation does not verify that repositoryUrl is a working browser URL or that defaultBranch matches the branch selected during cloning. Verify these values manually.

A successful validation of the first five rows does not prove that every repository in a larger CSV is accessible. Confirm that the personal access token has read access to every configured repository.

If repository access validation fails, check Git Server URL, repositoryName, Git Username, and Personal Access Token permissions. For more help, see Troubleshooting and FAQs.

Verify indexed content

After the initial crawl completes:

1. Search for a repository name.
2. Open a file result and confirm that its link opens the expected file in the Git server.
3. Open a directory result and confirm that its link opens the expected tree path.
4. Open a commit result and confirm that its link opens the expected commit.
5. Check a text or code file to confirm that its body is searchable.

If links or file bodies are incorrect, see Troubleshooting and FAQs.

Update the connector

To change the repositories or result-link values:

1. Update the repository CSV.
2. Upload the updated CSV in the connector configuration.
3. Save the connector.
4. Allow the next crawl to process the updated repository list.

Changing defaultBranch updates result URLs and branch metadata. It does not change the branch that Git Forge crawls.

To rotate the personal access token, replace Personal Access Token in the connector configuration and save it again. Ensure the replacement token has read access to every repository in the CSV.