◐ WIPE EFFECT
PING: -- ms
← Back to Articles

Tutorial 5 min read

Introducing Last Gate AI: A Portfolio Dashboard for Multi-Repo Shipping

How I keep active repos, project boards, roadmaps, and health checks close enough to answer one question: where should I work next?

I usually have several software projects alive at once. GitHub shows recent commits, the board holds the next tasks, and the host knows what is running. Before I can work, I have to remember how those pieces fit together for whichever app has gone quiet.

I built Last Gate AI to make that first pass quicker. It keeps the active portfolio in one dashboard, flags work that has gone stale or overdue, and opens the board or repository when I need to go deeper.

What Last Gate keeps together

An application can come from GitHub or be registered manually with a repository URL. Once it is in the portfolio, Last Gate can:

  • sync repository activity and available issue data
  • keep a Kanban board and roadmap with the app
  • surface stale syncs, overdue tasks, and active work
  • run intelligence checks when I need to recover context

The dashboard helps with triage. GitHub, the project files, and the deployment host remain the places I use to verify the underlying work.

Last Gate AI portfolio dashboard showing pinned apps, portfolio pulse metrics, and a lifecycle board for Development and Growth
Home triage: pinned apps, portfolio pulse, and a lifecycle board for what is in flight.

Step 1: Connect the services you use

The Settings wizard covers three integrations:

  • GitHub for repository sync, stack scans, and deployment detection
  • AI for optional live summaries and analysis agents
  • Vercel or Railway for optional deployment tracking

GitHub sync needs a personal token. The AI key is optional; without one, the analysis views use offline templates built from information the app already has.

Keys can live in .env.local or be entered in Settings. Restart the local server after changing the environment file.

Step 2: Add the apps that still matter

From Applications, import a GitHub repository or register one manually. I avoid importing every abandoned experiment because it turns the attention list into background noise.

For each active app:

  • open the Overview tab
  • run Sync now
  • check the last commit, recent activity, and available issue data

Last Gate stores the synced snapshot so the home dashboard can find stale work without opening every repository. GitHub remains authoritative for the repository and its activity.

Step 3: Triage from home

The home dashboard groups the signals I need at the start of a session:

  • Needs attention for stale repos, missing syncs, and overdue tasks
  • Work in flight for direct links to active boards and roadmaps
  • Refresh intelligence when a linked repository needs a new analysis

I read the attention list first and open the project that is actually holding something up. That replaces the usual round of checking GitHub, a board, and a deployment panel before deciding where to begin.

Step 4: Keep the board simple

Each application has a Kanban board with three basic lanes:

  • Backlog
  • In progress
  • Done

Tasks support priorities and due dates. An overdue item on the dashboard links straight to its task.

Use the Board | Roadmap switcher in the app header when dates matter more than columns. I use Roadmap for releases and milestones, and Board for the daily work. Done means deployed or otherwise finished for that project, not merely merged.

Last Gate AI Kanban board for last-gate-ai with Backlog, To Do, In Progress, and Done columns plus product lifecycle guidance
Per-app board: keep lanes simple, respect WIP limits, and advance the product lifecycle when the work is actually ready.

Step 5: Run intelligence when context is thin

The Intelligence tab can run a complete analysis or one individual check:

  • stack
  • architecture
  • AI summary
  • security
  • headroom
  • deployments

I run the full pass after a large sync or before a release. The output gives me a checklist, but I still read the diff, review the board, and verify the deployment. Without an AI key, offline templates produce structured results from the data already available to the app.

Last Gate AI Intelligence tab showing a completed seven-step analysis and repository stack scan for Next.js, React, Prisma, and TypeScript
Intelligence after a full pass: stack, architecture, summary, security, headroom, and deployments in one place.

How it fits into my week

  • In the morning, I clear overdue and stale items before opening the other tools.
  • During the week, I sync apps with new commits and open Roadmap only when dates need attention.
  • Before a release, I run the full intelligence pass and verify the security and headroom notes by hand.
  • After a large import, I pin the active apps and leave the rest quiet.

Technical notes

Last Gate AI is a Next.js App Router project with Prisma and a local SQLite database in development.

For a personal operations dashboard, that setup gives me:

  • fast local iteration
  • simple migrations
  • no cloud database tax while you are still shaping the workflow

GitHub sync uses a personal token. AI summaries go through Vercel AI Gateway when a key is present, and host integrations add deployment tracking.

To run it locally:

  • npm install
  • npx prisma migrate dev
  • npm run dev

Open localhost:3000, add the integrations you need, and start with one active app.

Boundaries

  • Read diffs and review pull requests in the repository.
  • Keep CI and host controls in the systems responsible for them.
  • Import the projects you still maintain instead of every repository in the account.

Last Gate gives me a short route from a portfolio-level warning to the board, repository, or deployment that needs work. Try the live dashboard from the Last Gate AI development page, or open Last Gate AI directly.

VIEW MORE WORK