AGENTS / GITHUB / Owlynn
githubinferredactive

Owlynn

provenance:github:maThiaslI152/Owlynn
WHAT THIS AGENT DOES

Owlynn is a local AI agent designed to boost productivity by acting as a helpful coworker. It runs entirely on your computer, ensuring privacy and offline functionality. The agent uses a combination of two language models and a sophisticated tool system to handle various tasks. Users can leverage its skills for research, document creation, task management, and more. Owlynn's architecture includes LangGraph for orchestration, a local search engine, and a memory system to retain information. It's ideal for developers, researchers, or anyone seeking a powerful, private AI assistant.

PROBLEM IT SOLVES

Owlynn solves the problem of juggling multiple tasks and needing assistance with research, writing, and organization. Instead of manually searching the web, creating documents, or managing to-do lists, users can rely on Owlynn to automate these processes and provide intelligent support, saving time and increasing efficiency.

View Source ↗First seen 2mo agoNot yet hireable

CAPABILITIES & CONSTRAINTS

TECH & STACK
pythonlanggraphllmlocal-aiproductivitytaurifastapichromadb
README
# Owlynn — Local AI Cowork Agent

A private, local-first AI productivity agent inspired by Anthropic's Cowork. Runs entirely on your machine with LangGraph orchestration, dual local LLMs, and a Cowork-style tool system.

## Quick Start

```bash
# 1. Start LM Studio with both models loaded (port 1234)
# 2. Start services + backend + desktop app:
./start.sh
```

## Architecture

- **Orchestrator**: LangGraph (stateful cyclic graph with 7 nodes)
- **Small Model**: Liquid LFM2.5-1.2B (routing + quick answers, ~700 tok/s)
- **Large Model**: Qwen3.5-9B (reasoning + tool calling, ~9 tok/s)
- **Backend**: FastAPI + WebSocket streaming
- **Frontend**: Tauri desktop app (offline-capable, vendored dependencies)
- **Search**: SearXNG (self-hosted) → DDG/Bing fallbacks
- **Memory**: Mem0 + ChromaDB + JSON files

## Tools (20)

| Category | Tools |
|----------|-------|
| Web | web_search, fetch_webpage |
| Files | read, write, edit, list, delete workspace files |
| Documents | create_docx, create_xlsx, create_pptx, create_pdf |
| Compute | notebook_run, notebook_reset |
| Memory | recall_memories |
| Tasks | todo_add, todo_list, todo_complete |
| Skills | list_skills, invoke_skill |
| HITL | ask_user (with choice buttons) |

## Skills (11)

Reusable prompt templates in `skills/`. Zero token cost until invoked.

| Skill | Triggers |
|-------|----------|
| Research Assistant | research, investigate |
| Document Summarizer | summarize, tldr |
| Morning Briefing | briefing, daily summary |
| Visual Comparison | compare, vs, chart |
| Data Visualization | graph, plot, histogram |
| Meeting Notes | meeting notes, action items |
| Email Drafter | draft email, compose |
| Report Generator | create report, weekly report |
| Presentation Builder | make slides, powerpoint |
| Content Rewriter | rewrite, polish, proofread |
| Brainstorm | brainstorm, ideas, what if |

## Prerequisites

- Python 3.12+
- LM Studio with both models loaded
- Podman (for ChromaDB, SearXNG containers)
- Rust & Cargo (for Tauri desktop app)

## Documentation

- [Architecture Overview](docs/ARCHITECTURE_OVERVIEW.md)
- [Agent Flow](docs/AGENT_FLOW.md)
- [Tools Reference](docs/TOOLS.md)
- [Chat Protocol](docs/CHAT_PROTOCOL.md)
- [API Reference](docs/API_REFERENCE.md)

## Project Structure

```
src/agent/       LangGraph orchestration (graph, nodes, state, LLM pool)
src/api/         FastAPI backend (server, file processor)
src/memory/      Memory system (Mem0, profiles, topics, interests)
src/tools/       Tool implementations (web, files, docs, notebook, todo, skills)
src/config/      Settings and M4 optimization config
frontend/        HTML/CSS/JS + vendored dependencies
skills/          Reusable prompt templates (markdown)
data/            User data (profile, memories, todos, topics)
tests/           Test suite
docs/            Architecture and API documentation
```

PUBLIC HISTORY

First discoveredMar 28, 2026

IDENTITY

inferred

Identity inferred from code signals. No PROVENANCE.yml found.

Is this yours? Claim it →

METADATA

platformgithub
first seenMar 17, 2026
last updatedMar 27, 2026
last crawled2 months ago
version

README BADGE

Add to your README:

![Provenance](https://getprovenance.dev/api/badge?id=provenance:github:maThiaslI152/Owlynn)