Module Architecture
Logical modules, capability services, and infrastructure dependencies in the Monkeys ecosystem
This document describes the logical module architecture of the Monkeys ecosystem. It is not a deployment guide.
For local startup and deployment-oriented instructions, see LOCAL_DEVELOPMENT.md, AGENT_BOOTSTRAP.md, and the compose files under docker/.
Layered View
Core Product
The core product is the minimum product surface for the Monkeys platform:
| Module | Repository | Responsibility |
|---|---|---|
| Studio frontend | monkeys-studio | Main user-facing Studio workspace. |
| Kernel frontend | monkeys-kernel | Platform and kernel management frontend. |
| Compute frontend | monkeys-compute | Compute control-plane frontend. |
| Main backend | monkeys-server | Core API, workflow entrypoint, routing, marketplace, and service bus. |
Capability Services
Capability services extend the core product and are deployed only when their feature area is needed.
| Capability | Repositories | Relationship |
|---|---|---|
| Agent runtime | monkeys-agent-server, monkeys-mcp-server, monkey-tools-agentkits | Adds agent runtime, MCP access, and reusable agent tools. |
| Billing and metering | monkeys-billing-server | Tracks usage, pricing, balances, and billing records. |
| Data workflows | monkeys-data-server, monkeys-conductor-worker | Adds data service APIs and workflow task execution. |
| Webhooks | monkeys-webhook-server | Handles outbound notifications and external webhook channels. |
| Third-party tools | monkey-tools-third-party-api, monkey-tools-third-party-api-python, monkey-tools-sandbox, monkey-tools-agentkits | Adds external API adapters, Python tools, code sandboxing, and agent toolkits. |
| Image and ComfyUI flows | monkeys-comfyui-worker | Executes ComfyUI/image generation queues. |
| Training flows | monkeys-training-worker | Executes model training and model-test background tasks. |
Infrastructure Dependencies
Infrastructure is shared and lives under docker/infrastructure/ for local use.
| Infrastructure | Used by |
|---|---|
| PostgreSQL | Main backend, agent, billing, data, webhook, training worker, Conductor |
| Redis | Main backend, workers, queues, cache, background dispatch |
| Elasticsearch | Conductor indexing and search-related flows |
| Conductor | Workflow orchestration and task dispatch |
| MinIO | Local object storage for files and assets |
| Prometheus / Grafana | Optional local metrics and dashboards |
Deployment Shape
Monkeys services are deployed on demand. The local services compose uses profiles to reflect that model.
Ownership Rules
monkeysowns ecosystem documentation, local infrastructure compose, service compose, and agent bootstrap guidance.- Each application or service repository owns its own source code, Dockerfile, config examples, tests, and build commands.
- README files describe repository relationships. Startup and deployment details belong in
LOCAL_DEVELOPMENT.md,AGENT_BOOTSTRAP.md, anddocker/*/README.md. - Feature work must be tested in the owning repository; run local test and build commands successfully before pushing.