Guard Deployments: Browserbase + OpenAI Regression Diagnosis

Guard Deployments: Browserbase + OpenAI Regression Diagnosis

Catch deployment regressions automatically: this n8n workflow receives a deployment webhook (or runs manually), executes a smoke-test plan in demo or via Browserbase, and uses OpenAI to generate a structured regression diagnosis—then deduplicates and files a GitHub issue while notifying Slack (and optional email).

What this workflow does

• Triggers on a POST deployment webhook or a manual run, loading the workflow configuration and a default TEST_PLAN.
• Normalizes deployment payloads to extract the target URL, environment, version, commit, branch, and provider.
• Runs test plan items by iterating through the JSON plan:
  • Generates demo mode results, or
  • Runs live browser-based checks through the Browserbase API and collects test logs.
• Aggregates results into a single report and detects failures.
• If failures occur, classifies issues and sends evidence to an OpenAI-backed AI agent to return a structured regression diagnosis (with a basic report fallback if the AI step fails).
• Creates incident records safely by generating a regression fingerprint, searching for an existing matching GitHub issue, and creating a new issue only when none exists.
• Communicates outcomes by posting the incident summary (and GitHub link) to Slack and optionally emailing when tests fail; otherwise it sends a Slack success notification.

Use cases

• SaaS teams that want automated regression diagnosis immediately after each deployment.
• Automation engineers running smoke-test plans that combine demo validation with real Browserbase checks.
• Operations workflows that need GitHub issue deduplication to prevent alert fatigue.

Technical details

• Integrations: Browserbase, OpenAI, GitHub, Slack (optional email).
• n8n nodes/logic: if, set, code, merge, slack, github.
• Key configuration: APPLICATION_URL, TEST_EMAIL, TEST_PASSWORD, TEST_PLAN, GITHUB_REPO/SLACK_CHANNEL, and DEMO_MODE.