Laufwise Platform

A governed agent runtime: the runtime that lets you run any agent inside an enforced, auditable process contract grounded in your real systems of record. See PLATFORM_PLAN.md for the thesis, architecture, and roadmap. CI/CD is documented in docs/CICD.md.

This repo is a monorepo with a clear backend/frontend split:

laufwise_platform/
├── backend/    # FastAPI — the control plane API, providers, workloads, infra layers
├── frontend/   # Next.js — the Studio (agents, calls, connections, workflows) and the runs console
├── PLATFORM_PLAN.md
└── CLAUDE.md   # engineering rules for this repo

The control plane itself is the laufwise primitive, consumed as a dependency by the backend (backend/app/control_plane/). The platform wraps it with an HTTP API, identity, memory, observability, and the agent surfaces ("workloads").

Quick start

Backend and frontend run independently. See each folder's README.md.

# backend
cd backend && python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
uvicorn app.main:app --reload          # http://localhost:8000  (docs at /docs)

# frontend (in another terminal)
cd frontend && npm install
npm run dev                            # http://localhost:3000

Or with Docker: docker compose up.

Architecture (one screen)

WORKLOADS (agents) — conversational | back-office task | document/workflow   ← peers
        │  every action passes through ▼
CONTROL PLANE (= laufwise)
  precond → tool allowlist → approval → execute → postcond → checkpoint + trace
        │            │              │              │
   IDENTITY      MEMORY        STATE (SoR)     OBSERV / AUDIT

The workload never calls tools directly — every action routes through a runbook step the runtime enforces against the real system of record.