Build AI agents that keep running — across your home, robots, and spare computers.
Docs | Get started | Architecture | Home Assistant Addon | Issues
Wactorz is a self-hosted runtime for AI that needs to do more than answer once. Describe an automation in chat and its planner can write a Python agent, run it as a supervised actor, and show you what it is doing in a live dashboard. Agents can listen to sensors and MQTT topics, act through Home Assistant or other services, keep state across restarts, and run on another machine when the job belongs closer to the hardware.
Bring a hosted model from Anthropic, OpenAI, Gemini, or NVIDIA NIM, or keep model inference local with Ollama. The runtime, agent state, dashboard, and MQTT control plane remain on hardware you control.
https://github.com/user-attachments/assets/e832eed9-ba61-4872-8c20-5d0e8cfe14d5
Who it is for
Wactorz is built first for technical hobbyists and makers: people connecting a smart home, robot, camera, Raspberry Pi, or old laptop who are comfortable using a terminal and editing a configuration file. It is also useful to Python developers who want an extensible actor runtime instead of a one-shot agent script.
It is currently a beta, not a no-code consumer app or a multi-user hosted service. Agents can execute generated Python, so use it on systems you control and review the security model before exposing it to a network.
What you can build
- Persistent automations that react to sensor readings, schedules, and MQTT events.
- Home Assistant actions described in chat rather than hand-authored for every case.
- Robot and edge-device agents that run close to their cameras, microphones, or motors.
- Small groups of agents that keep state, restart independently, and move between nodes.
- Custom Python agents with REST, MCP, Discord, Telegram, or WhatsApp as an interface.
Quick start
The most predictable first run uses Docker Compose. You need Git, Docker with the Compose plugin, and an API key for your chosen model provider.
git clone https://github.com/waldiez/wactorz
cd wactorz
cp .env.template .env
Open .env and set at least these values:
MQTT_PASSWORD=choose-a-long-random-password
LLM_PROVIDER=anthropic
LLM_MODEL=claude-sonnet-4-6
LLM_API_KEY=your-key-here
Then start Wactorz:
docker compose --profile python up -d
Open the dashboard at http://localhost:8888. Follow the
logs with docker compose logs -f wactorz-python and stop everything with
docker compose --profile python down.
On Windows, use Copy-Item .env.template .env instead of cp. To install from
PyPI, use Ollama, or run inside Home Assistant, see the
deployment guide. Repository
contributors should start with the
development setup.
Use it as a library
Bring an agent you already have — a trained model, a class with its own loop — and Wactorz supervises it, restarts it, persists its state and shows it on the dashboard. One decorator is enough:
import wactorz
@wactorz.agent(subscribes="sensors/imu/#", publishes="anomalies/imu")
def detect(reading: dict) -> dict | None:
return reading if MODEL.score(reading) > 4.0 else None
wactorz.run(agents=[detect], minimal=True) # monitor + dashboard + your agent, no LLM needed
# or, inside a program that already has an event loop: await wactorz.serve(agents=[detect], minimal=True)
Or name it in WACTORZ_AGENTS=mypkg.agent:detect for a normal wactorz start, or list it
as a wactorz.agents entry point in your package and name that package in WACTORZ_EXTENSIONS. wactorz.pipeline(...) groups several
such agents with a schedule and rules, wiring checked up front. The Home Assistant agents start only
when Home Assistant is configured. See Using Wactorz as a library
and examples/imu_anomaly/.
Try an automation
Once the dashboard is connected to the services or devices named in a request, try:
when a person is detected in my pc camera, open the office light
when the door opens, make reachy wakeup
when the light has been on for too long, send me a discord notification
Architecture
flowchart LR
User["User<br/>CLI, REST, Discord, Telegram, HA"] --> Main["MainActor<br/>intent routing"]
Main --> Actuate["OneOffActuatorAgent<br/>direct service calls"]
Main --> Planner["PlannerAgent<br/>pipeline planning"]
Main --> HA["HomeAssistantAgent<br/>REST + WebSocket"]
Main --> Chat["LLM reply<br/>streaming response"]
Planner --> Dynamic["DynamicAgents<br/>LLM-generated runtime code"]
Actuate --> Bus["MQTT broker"]
HA --> Bus
Dynamic --> Bus
Bus --> Dashboard["Live dashboard<br/>agents, logs, cost, heartbeats"]
Bus --> Remote["Remote nodes"]
Bus --> External["Sensors, services, and IoT systems"]
Interfaces
| Interface | How to use it |
|---|---|
| CLI | python -m wactorz |
| Live dashboard | http://localhost:8888 |
| REST API | python -m wactorz --interface rest |
| Discord | python -m wactorz --interface discord |
| Telegram | python -m wactorz --interface telegram |
python -m wactorz --interface whatsapp |
|
| MCP server | wactorz-mcp |
| Home Assistant addon | One-click install inside the HA Supervisor |
Choose a model
Set LLM_PROVIDER, LLM_MODEL, and LLM_API_KEY in .env. The generic key works
with every hosted provider, or you can use its provider-specific variable.
| Provider | LLM_PROVIDER |
Example model | Provider-specific key |
|---|---|---|---|
| Anthropic | anthropic |
claude-sonnet-4-6 |
ANTHROPIC_API_KEY |
| OpenAI | openai |
gpt-4o |
OPENAI_API_KEY |
| Google Gemini | gemini |
gemini-2.5-flash |
GEMINI_API_KEY |
| NVIDIA NIM | nim |
meta/llama-3.3-70b-instruct |
NIM_API_KEY |
| Ollama | ollama |
llama3 |
No key; configure OLLAMA_URL |
Wactorz logs the resolved provider and model at startup. Advanced setups can
route different jobs to different models
and use the evaluation harness
before choosing them. Optional LLM_TEMPERATURE, LLM_MAX_RETRIES, and
LLM_TIMEOUT_S settings are documented in .env.template.
Security
Wactorz is under active development. The perimeter is closed by default; the remaining caveats below are about what an authenticated caller can do.
What protects an install:
- The API and dashboard require a key. Set
API_KEYand every route, the WebSocket handshake, the Prometheus scrape, and the login flow are authenticated — constant-time comparison, session cookies that survive a restart, and sign-in throttling. - The server binds to
127.0.0.1. Reaching it from the network is deliberate: setWACTORZ_BIND_HOST, and startup refuses unlessAPI_KEYis set too, orWACTORZ_EXPOSED_OK=1says the only way in is already authenticated. A key short enough to guess is warned about. - The broker requires credentials. Anonymous MQTT is off, and remote nodes are given credentials rather than connecting openly.
- Origin and Host allow-lists guard the HTTP surface and the WebSocket handshake against cross-site requests and DNS rebinding.
What to still assume:
- Agents execute code. The planner generates and runs Python, and remote nodes run code delivered over MQTT. Anyone holding the API key or the broker credentials can run code on the host and on every connected node — treat both as root-equivalent, the same way you would an Ansible control node.
- Generated code is screened by a best-effort blocklist, not a sandbox.
Deployment rules:
- ✅ Set
API_KEYbefore exposing anything beyond loopback. - ✅ Prefer the Home Assistant add-on, which keeps the UI behind HA's ingress auth.
- ✅ Keep the broker on a network you control, with credentials set.
- ❌ Do not run it as a multi-user or multi-tenant service — there is one key, not per-user accounts, and no isolation between agents.
- If you reach it remotely, prefer a VPN or an authenticating reverse proxy over a bare port-forward, even with a key set.
More detail in docs/security.md. Found a security issue? Please see SECURITY.md rather than opening a public issue.
Repository Map
| Path | What lives there |
|---|---|
wactorz/ |
Python actor runtime, built-in agents, interfaces, monitoring, HA integration |
frontend/ |
Vite + TypeScript card dashboard |
ha-addon/ |
Home Assistant Supervisor addon |
docs/ |
Markdown docs source |
infra/ |
Mosquitto, Prometheus, nginx, and HA configs |
tests/ |
Python test suite |
Documentation
| Start here | For |
|---|---|
| Installation | Install with pip, configure, upgrade |
| Windows | Windows and ARM64 setup |
| Docker Hub | Run from Docker without cloning the repo |
| Architecture | Actor system, supervision, MQTT flow |
| Agents | Built-in agents, recipes, and dynamic agents |
| Library | Your own agents, pipelines and models, with examples |
| Pipelines | Reactive automation patterns |
| Remote nodes | Edge deployment over SSH |
| Interfaces | CLI, REST, chat platforms, dashboard, MCP |
| API reference | REST endpoints and payloads |
| Deployment | Docker, Home Assistant add-on, environment setup |
| Prometheus | Metrics and monitoring |
| Security | Auth, exposure, broker credentials, threat model |
| Evaluation harness | Compare models per LLM call site |
| Technical reference | Deeper internals |
Contributors
|
Panagiotis Kasnesis 📆 💻 |
Lazaros Toumanidis 💻 🎨 |
Chris 💻 📓 |
Amalia Contiero 💻 📣 |
Contributions of any kind are welcome. See CONTRIBUTING.md to get started.
Contributing
| What | How |
|---|---|
| Found a bug | Open an issue |
| Found an accessibility barrier | Tell us; ACCESSIBILITY.md says what we aim for and what we know falls short |
| Have an idea | Start a discussion |
| Want to code | Fork, branch, and open a PR against dev (main is releases only) |
| Docs, tests, UI | Same drill, open a PR |
| New agent recipe | Add it in wactorz/catalogue_agents/ and open a PR |
| Home Assistant | HA integrations and addon config PRs are very welcome |
Read CONTRIBUTING.md for setup instructions, code style, and the PR process.
License
Apache 2.0. Free to use, modify, and distribute.
Metadata
Release files for wactorz 0.8.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| wactorz-0.8.0.tar.gz | 6.6 MB | With provenance | Details for wactorz-0.8.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| wactorz-0.8.0-py3-none-any.whl | Python 3 | none | any | Details for wactorz-0.8.0-py3-none-any.whl |
Total release size: 9.1 MB
Release files / wactorz-0.8.0.tar.gz
| Download URL | wactorz-0.8.0.tar.gz |
|---|---|
| Size | 6.6 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
690bf33b14f8bbafe145cd81975388bd2ad823c59f2e4f032785f98dbd49e43b
|
|
BLAKE2b-256 checksum How to use checksums |
1c48e19e6696c651418022666c086525bc96091678c72c568d7d6bf8190f78e6
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 15, 2026.
Transparency logRelease files / wactorz-0.8.0-py3-none-any.whl
| Download URL | wactorz-0.8.0-py3-none-any.whl |
|---|---|
| Size | 2.4 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
278526aed512196dadb749dd9324c42bfe8cfa23a9c73880f7b1afb6fdd2341c
|
|
BLAKE2b-256 checksum How to use checksums |
fcd44e117591c5a7e270e0c8683af4b37db75fc2076247389cb140185cb13b30
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 15, 2026.
Transparency logSecurity for wactorz 0.8.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for wactorz-0.8.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 15, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.