openyfai-aria 1.0.3

Local-first AI agent with memory, knowledge graph, tools, web dashboard, and Telegram (OpenYF AI).

ARIA by OpenYF AI

ARIA is a local AI agent that lives on your machine, remembers everything, and works 24/7 — even while you sleep.

Built by OpenYF AI as part of an open journey toward general intelligence.

Install

pip install openyfai-aria

Start

aria setup

That's it. Open http://127.0.0.1:8000 and start talking. No options A and B. No jargon. One path.

Developing from source (contributors)

Clone the repo and use an editable install when you are changing Python or the Next.js UI:

git clone https://github.com/openyfai/aria.git && cd aria
pip install -e ".[full,dev]"

If you change the web frontend, build it once so aria web can serve aria-ui/out from your checkout:

cd aria-ui && npm install && npm run build && cd ..

Release builds (CI) copy this output into aria/web_dist/ for PyPI wheels — see CONTRIBUTING.md.

Why ARIA

• Local-first memory and state backed by SQLite.
• A visible graph of what the agent believes.
• Provider and model choice from the setup wizard or aria setup.
• Operator controls for approvals, usage, Telegram pairing, and remote access.
• Markdown skills that let contributors extend behavior without editing Python.

More ways to run

Path	When to use
Web UI (aria web)	Default: finish setup in the browser.
Terminal agent (aria after setup)	Keyboard-first use.
CLI setup only (aria setup)	You prefer not to use the web wizard once.
Docker	Deploy to a server or share on a LAN; needs .env and a web API key. See docs/quickstart.md.

Requirements

• Python >=3.11 (matches package metadata on PyPI).
• Node 20+ — only if you develop or rebuild aria-ui/ from source (not required for standard pip install wheels).
• Docker — optional.

Safe defaults

• Tool approvals are on.
• Terminal execution is off.
• Direct code writes are off.
• Background actions are off.
• Remote web binds require ARIA_WEB_API_KEY.
• Telegram is deny-by-default until paired from Operator.

Read SECURITY.md before exposing ARIA beyond localhost.

Power user / troubleshooting

If chat or setup fails after you have saved provider keys, see docs/quickstart.md for aria doctor, aria doctor --ping, and related commands. You do not need these for a first successful run.

Provider support

Gemini, OpenAI, Anthropic, OpenRouter, DeepSeek, Mistral, Groq, Ollama, and local OpenAI-compatible endpoints. Settings come from the wizard, aria setup, data/ runtime files, or environment variables.

Skills

Add a Markdown file under skills/, restart ARIA, and the agent can use that workflow. Start with skills/repo_researcher.md or skills/README.md.

Repository guide

• Core loop: aria/core/cognitive_loop.py
• Providers: aria/llm/
• Web backend: scripts/web_server.py
• CLI: scripts/cli.py
• Telegram: scripts/telegram_bot.py
• Web UI: aria-ui/src/
• Packaged static UI (release builds): aria/web_dist/
• Skills: skills/

Development

python -m pytest tests
ruff check .
cd aria-ui && npm install && npm run lint && npm run build

Project docs

• docs/quickstart.md — install paths, Docker, troubleshooting
• CONTRIBUTING.md
• SECURITY.md
• CODE_OF_CONDUCT.md
• docs/planning/ROADMAP.md — phased roadmap
• docs/planning/launch-checklist.md — pre-public-release checklist
• docs/planning/launch-issue-backlog.md — seeded “good first issue” ideas
• CHANGELOG.md

License

Apache-2.0. See LICENSE.