Bulk Removing GitLab Webhooks with Python

We were retiring an old integration and needed to clean up the associated webhooks from all our GitLab projects. The problem: hundreds of projects, each potentially with multiple webhooks, all added over the years.

The GitLab UI lets you manage webhooks one project at a time. That wasn’t going to work.

The approach

The GitLab API has full CRUD for webhooks. The plan was straightforward:

1. Paginate through all projects I’m a member of
2. For each project, list its webhooks
3. Delete any webhook whose URL contains the target keyword

""" Bulk-delete GitLab webhooks matching a keyword """
import requests

GITLAB_TOKEN = '<your_access_token>'
GITLAB_API_URL = 'https://gitlab.com/api/v4/'

KEYWORD = 'your-integration-keyword'

headers = {'Private-Token': GITLAB_TOKEN}

def get_all_projects():
    projects = []
    page = 1
    while True:
        response = requests.get(
            f"{GITLAB_API_URL}projects?page={page}&per_page=100&membership=true",
            headers=headers
        )
        if response.status_code != 200:
            break
        data = response.json()
        if not data:
            break
        projects.extend(data)
        page += 1
    return projects

def get_project_webhooks(project_id):
    response = requests.get(f"{GITLAB_API_URL}projects/{project_id}/hooks", headers=headers)
    return response.json() if response.status_code == 200 else []

def delete_project_webhook(project_id, hook_id):
    requests.delete(f"{GITLAB_API_URL}projects/{project_id}/hooks/{hook_id}", headers=headers)

def main():
    projects = get_all_projects()
    for project in projects:
        webhooks = get_project_webhooks(project['id'])
        for webhook in webhooks:
            if KEYWORD in webhook['url']:
                print(f"Removing {webhook['url']} from {project['name']}")
                delete_project_webhook(project['id'], webhook['id'])

if __name__ == "__main__":
    main()

A few things worth noting

The membership=true filter is important — without it you’d get every public project on GitLab. Pagination is essential; the API caps responses at 20 projects by default and won’t tell you there’s more unless you keep requesting.

The script is intentionally simple: no concurrency, no dry-run flag, no progress bar. I ran it once, watched the output, and it finished in under a minute across the full project list.

What I’d add for a larger scale

If this needed to run regularly or against a much larger instance:

Addition	Why
Dry-run mode	Print matches before deleting
Rate limiting	GitLab will push back if you hit it too fast
Parallel requests	Speed up across hundreds of projects

For a one-time cleanup job, none of that was necessary. The script did exactly what was needed and nothing more.