Every operation must be done on its own branch. Never commit directly to main.
<type>/<short-description>
| Type | Used For | Example |
|---|---|---|
feature/ |
New feature or enhancement | feature/nextjs-dashboard |
bugfix/ |
Fixing a non-urgent bug | bugfix/moe-router-edge-case |
hotfix/ |
Urgent production fix | hotfix/kafka-connection-leak |
release/ |
Release preparation | release/v1.2.0 |
chore/ |
Refactoring / cleanup / housekeeping | chore/gitignore-update |
docs/ |
Documentation only | docs/architecture-diagram |
test/ |
Adding or fixing tests | test/integration-kafka |
perf/ |
Performance improvements | perf/moe-scoring-vectorize |
infra/ |
Infrastructure / DevOps changes | infra/helm-chart-agents |
lang/ |
New language service (Rust, TS, etc) | lang/rust-kafka-consumer |
- One branch = one concern. Don't bundle unrelated changes.
- Branch from
mainunless it's a hotfix on production. - Never force-push to
main. - Merge via PR (or explicit user approval in this project).
- Delete branch after merge.
<type>(<scope>): <short description>
[optional body]
[optional footer: refs #issue]
feat(moe): add ensemble routing for low-confidence tasks
fix(kafka): handle reconnect on broker restart
chore(deps): bump pydantic to 2.6.0
test(moe): add scoring edge case for zero vectors
docs(arch): update kafka topic schema table
infra(k8s): add KEDA ScaledObject for engineer-backend
lang(rust): add fast cosine similarity service
| Branch | Type | Status | Description |
|---|---|---|---|
main |
— | Stable | Production base |
chore/git-workflow-setup |
chore | Active | Git conventions + .gitignore |
Update this table every time you create or merge a branch.
To maintain a clean and reliable history, we follow a strict Pull Request (PR) workflow:
- Create Feature Branch: Use the standard
type/descriptionnaming. - Commit with Scope: Use conventional commits (e.g.,
feat(agents): ...). - Raise PR: Push your branch to GitHub and open a Pull Request against
main. - Pass Checks: Ensure CI/CD passes for Go, Python, and Rust.
- Merge: Once approved, merge via PR. Do not commit directly to
main.
Proximus is a polyglot system designed for performance and scalability:
- Go 1.24.0 — API Gateway, Orchestrator, health monitoring, and system-level microservices.
- Python 3.12 — AI Agents (Amazon Bedrock/Nova), TUI, and observability integrations.
- Rust — High-performance scoring (MoE) and security-critical services (AST validation).
- TypeScript / Next.js 14 — Real-time analytics dashboard and React Flow visualization.
- Apache Kafka — Main event bus for agent-to-orchestrator communication.
- PostgreSQL 15 & Redis 7 — Structured data persistence and low-latency caching/PubSub.
- Docker & Terraform — Containerization and Infrastructure as Code.