How I Shipped 15 Green MRs to GitLab in One Day: Inside Our Modern Contributor Tech Stack

Описание и применение

Команда GitLab успешно отправила 15 мердж‑реквестов за один день, сохранив 100 % зеленый статус CI и назначив всех ревьюеров, благодаря новому трёхуровневому стеку с MCP, LSP и AI‑Gateway.

Задача

Ship multiple upstream merge requests across Ruby, TypeScript, and Python in a single sprint while keeping CI pipelines green and reviewers assigned.

Результат

All 15 merge requests passed CI with 100% green status, were assigned to reviewers, and contributed 327 points, bringing the team to 4,197 live contributor points and approaching the Level 4 milestone.

Рабочий процесс

• 1. Implemented 9 MCP tool endpoints in GitLab Core to expose CI schedules, trigger pipelines, and manage work items.
• 2. Fixed critical WebSocket and UTF‑8 handling bugs in the GitLab LSP to ensure stable AI interactions.
• 3. Added idempotency guard and exception mapping in the AI Gateway to prevent infinite loops and improve error handling.
• 4. Ran local pre‑flight checks (Danger, RuboCop, predictive RSpec) before opening MRs.
• 5. Assigned all MRs to maintainers and domain reviewers immediately after creation.
• 6. Monitored CI success rate and reviewer coverage to ensure 100% green and 100% assigned status.

Доказательства

15 out of 15 Merge Requests 100% Green in CI, 100% assigned to maintainers and domain reviewers

Источник

We implemented and polished 9 distinct MCP tool endpoints

Источник

WebSocket Upgrade Rejection Ping Leak ("!4537", "#3364")

Источник

Identical Tool-Call Guard Enabled by Default ("!7641", "#3394")

Источник

Run Danger, RuboCop, and predictive RSpec locally before opening an MR

Источник