strata-ai-cli 0.1.3

Strata AI CLI. Scaffold Enterprise ready AI agents, APIs & ML Pipelines.

Strata AI CLI

Thin scaffolding orchestrator. Generates Day 1 production repos. Enforces platform/dev boundaries.

📦 Installation

uvx strata-ai-cli  # no install required
# or pip install strata-ai-cli

🛠️ Commands

Command	Purpose	Output
create-agent	ReAct, HITL, Orchestrator, etc.	agent/, evals/, deployment/, tests/, pre-wired app/main.py
create-api	Generic FastAPI microservice	services/, routers/, repositories/, task polling
create-batch	Spark/Kubeflow/Airflow pipelines	pipelines/, notebooks/, io/, async queue hooks
create-serving	MLflow/Cloud Run/GKE model serving	serving/, terraform/, cloudbuild.yaml, auth specs

📐 Template Contract

All commands share a single base template. Only app/main.py router inclusion and domain folder differ:

uvx strata-ai-cli create-agent my-agent
cd my-agent && uv run dev  # identical bootstrap for agent/api/batch

Platform vs Developer Ownership

PLATFORM OWNS (dev no-touch)          DEV OWNS
app/core/lifespan.py                  app/schemas/
app/core/middleware.py                app/services/
app/core/exceptions.py                app/routers/
app/core/dependencies.py              app/repositories/
app/routers/agent.py (pre-generated)  agent/agent.py
                                      agent/tools.py
                                      agent/config.py
                                      agent/prompts/

🔄 Adding Commands

Follow the 4-step pattern in CONTRIBUTING.md. No changes to existing commands required. Jinja2 templates receive standard context (project_name, sdk_version, etc.).

🚦 Day 1 Workflow

1. uvx strata-ai-cli create-<type> <name>
2. cp .env.example .env && uv sync
3. uv run dev → hot-reload :8080, console OTel, mock adapters, RFC 9457 errors
4. uv run test → deterministic, MockRuntime injected, ≥80% coverage enforced