Preview environments and code review, run on your own infrastructure.

DevPlatform spins up an ephemeral environment for every pull request, then hands it to an AI reviewer, a doc generator, and a test writer — the way Vercel and Netlify do it, except it runs in your cluster and your logs never leave the building.

View on GitHub Read the quick start
Go 1.22 · Gin Next.js 14 Apache 2.0

What it replaces

The preview-environment workflow, without handing your code to someone else's cloud

Most teams get ephemeral environments from a hosted platform, and AI code review from a separate bolt-on. DevPlatform runs both from one control plane, on infrastructure you already operate.

01

Self-hosted, GitOps-driven

Every PR gets a real kustomize overlay, deployed via ArgoCD — not a black-box build service.

02

AI wired into the pipeline

Review, docs, tests, and migration run as jobs against your own AI provider — OpenAI, Anthropic, or a local Ollama model.

03

Observable by default

Metrics, traces, and logs are wired up from the first `make dev` — not an integration you build later.

System design

One control plane, three lanes of work

The control plane (Go, Gin) sits between the frontend and everything else: it writes to Postgres directly, queues async work onto NATS, and leans on Redis for cache and live logs. Workers pick jobs off the queue; the whole path is instrumented and shipped to Grafana.

Frontend Next.js 14 Control plane Go · Gin PostgreSQL primary store NATS job queue Redis cache · live logs Workers async AI + GitOps jobs Prometheus metrics Tempo traces Loki logs Grafana dashboards

Capabilities

What's running under `make dev`

Six pieces of a normal platform team's stack, already wired together.

deploy

GitOps preview environments

A real environment on every pull request, rendered through kustomize and reconciled by ArgoCD.

ai

AI-assisted workflows

PR review, doc generation, test generation, and language migration, all running against a pluggable AI client.

org

Multi-project management

Group projects, environments, and deployments the way your team actually organizes work.

stream

Real-time updates

Deployment logs and status stream over WebSocket, so a build's progress shows up as it happens.

observe

Full observability

OpenTelemetry feeds Prometheus, Tempo, and Loki; Grafana ships with dashboards for all of it.

infra

Kubernetes native

Deploys with kustomize and staging/production overlays; ArgoCD is the recommended path to prod.

AI provider interface

Every AI job is one Go interface

DevPlatform ships an OpenAI, Anthropic, and Ollama client out of the box — and a mock provider for local dev. Add another by implementing AIClient in internal/ai/client.go.

type AIClient interface {
    Chat(ctx, messages) (string, error)
    ChatStream(ctx, messages) (<-chan string, error)
    ReviewPR(ctx, ReviewRequest) (*ReviewResult, error)
    GenerateDocs(ctx, DocsRequest) (*DocsResult, error)
    GenerateTests(ctx, TestsRequest) (*TestsResult, error)
    MigrateCode(ctx, MigrateRequest) (*MigrateResult, error)
}
ReviewPR
Code review with configurable focus

Point it at security, performance, or style, and it reviews the diff against that lens rather than everything at once.

GenerateDocs
Documentation generation

Drafts README updates, API docs, and architecture notes from the current state of the code.

GenerateTests
Test generation with a coverage target

Set the coverage goal you're aiming for; it writes tests toward that number.

MigrateCode
Cross-language migration

JS to TS, Python to Go, and similar moves — a starting translation you review, not a black box.

Get running

Local dev in one command

Everything — Postgres, NATS, Redis, the control plane, a worker, the frontend, and the observability stack — starts from a single Makefile target.

zsh~/devplatform
# clone and enter
git clone https://github.com/ATLURI0001/DevPlatform.git
cd DevPlatform

# copy and edit environment config
cp .env.example .env

# start every service
make dev
✓ postgres     ready on :5432
✓ nats         ready on :4222
✓ redis        ready on :6379
✓ control-plane listening on :8080
✓ frontend     ready on :3000
Frontendlocalhost:3000
Control plane APIlocalhost:8080
Grafanalocalhost:3001 (admin/admin)
Prometheuslocalhost:9090
NATS monitorlocalhost:8222
AI provider (.env)openai · anthropic · ollama · mock

Operating it

Built to run in a real environment, not just a demo

Security

  • auth JWT (HS256) with role-based access control
  • scope Permissions are scoped per project
  • secrets Kubernetes or Docker secrets — nothing in the repo
  • network Optional mTLS via service mesh
  • scanning Trivy, gosec, and CodeQL run in CI
  • supply chain SBOM via Syft, signed with cosign
  • updates Dependabot checks weekly

Observability

  • metrics Prometheus, visualized in Grafana
  • traces Tempo, via OpenTelemetry
  • logs Loki, also surfaced in Grafana
  • alerting Alertmanager, configured separately
  • dashboards Overview, pipeline, AI job performance, system health — included

Apache 2.0 — read the code, run it yourself.

Fork it, open an issue, or send a pull request. See CONTRIBUTING.md for the workflow.

View on GitHub Contributing guide