COMPASS
The open-source directory.
59 projects, communities, and foundations worth knowing. The ones we build on, and the ones worth joining. Every entry is open source.
AI Agent Frameworks
The stacks people use to build autonomous agents, from personal, memory-first assistants to production multi-agent systems.
Self-hosted autonomous AI agent with persistent memory, auto-generated skills, and multi-channel access.
Why: The reference stack for personal, memory-first agents you actually own — and the one this workshop runs.
Personal AI assistant that runs on your own devices and answers on the channels you already use.
Why: The flagship 'own your assistant' project — weekend hack to one of GitHub's most-starred repos.
Low-level orchestration for stateful, multi-agent workflows with cyclic graphs.
Why: The production-grade backbone for controllable, durable agent execution.
Orchestrates role-playing, collaborative autonomous agents ('crews').
Why: Fastest on-ramp to multi-agent collaboration.
The original goal-driven autonomous agent, now a platform for continuous agents.
Why: Launched the whole autonomous-agent category and still a huge community hub.
Microsoft's framework for multi-agent conversation and orchestration.
Why: Research-grade, big-backer framework driving multi-agent patterns.
TypeScript AI agent framework with a built-in observability console.
Why: Full code control plus production visibility for TS/JS builders.
Modern TypeScript framework for agents, workflows, and RAG.
Why: The batteries-included TS agent stack for the JS ecosystem.
Data framework connecting LLMs to your private data (RAG).
Why: The default toolkit for retrieval-augmented, knowledge-grounded agents.
Open-source LLM app platform with a visual agent/workflow builder.
Why: Ship LLM apps and agents without deep glue code.
Autonomous AI software-engineering agent (formerly OpenDevin).
Why: Leading open coding-agent that writes, runs, and debugs end to end.
Local LLM Inference
Run open models on your own hardware, one command to a private endpoint.
Run open LLMs locally with one command.
Why: The simplest, most popular on-ramp to local models.
High-performance C/C++ inference engine for LLMs on CPU/GPU.
Why: The foundational engine (and GGUF format) most local tooling is built on.
High-throughput, memory-efficient inference and serving engine.
Why: The production serving standard for scaling open models cheaply.
Desktop app to discover, run, and serve local LLMs with a GUI.
Why: Friendly local-model experimentation with an OpenAI-compatible endpoint.
Ship and run an entire LLM as a single executable file.
Why: Mozilla-backed — radically portable models with zero install.
Gradio web UI for running and chatting with local LLMs.
Why: The power-user hub for local model tinkering.
Open Models
Open-weight models worth knowing. Check the leaderboards monthly, this list moves fast.
Meta's family of open-weight foundation models.
Why: The most widely adopted open-weight base — anchor of the ecosystem.
Alibaba's open-weight LLM/multimodal family (Qwen3).
Why: Consistently top-tier open models across sizes, strong for local use.
Efficient open-weight models (incl. Mixtral MoE).
Why: European open-weight leader that punches above its parameter count.
Open-weight frontier-class reasoning and coding models.
Why: Redefined the open cost/performance frontier for reasoning models.
Nous Research's fine-tuned open models (Hermes series).
Why: Community-favorite instruct fine-tunes powering the Hermes agent stack.
Google DeepMind's lightweight open models from Gemini research.
Why: Big-lab open weights optimized for on-device deployment.
Systems & Languages
The languages and runtimes the safe, fast future is built on.
Memory-safe systems language with no garbage collector.
Why: Most-loved language for a decade and the safe-systems foundation for modern infra.
Simple, fast systems language and a drop-in C/C++ toolchain.
Why: Rising low-level contender prized for simplicity and C interop.
Concurrency-friendly compiled language for cloud and backend.
Why: Powers Docker, Kubernetes, and much of cloud-native infra.
All-in-one JavaScript runtime, bundler, and package manager.
Why: Dramatically faster JS toolchain reshaping the Node ecosystem.
Secure-by-default TypeScript/JavaScript runtime.
Why: Modern, secure runtime with first-class TS and web-standard APIs.
Fast, secure WebAssembly runtime from the Bytecode Alliance.
Why: Portable, sandboxed compute beyond the browser.
Web Frameworks
How the modern web gets built, fast by default.
Content-focused web framework with an islands architecture.
Why: Near-zero JS by default — and what this very site is built with.
Compiler-based UI framework that disappears at build time.
Why: Top-rated DX with a minimal runtime; SvelteKit for full-stack.
Fine-grained reactive UI library with no virtual DOM.
Why: Top-tier performance with React-like ergonomics.
Approachable, performant progressive framework for UIs.
Why: One of the largest independent frontend communities.
Databases
Where the data lives, from the phone in your pocket to analytics at scale.
The world's most advanced open-source relational database.
Why: The default serious database and base for a huge ecosystem.
Self-contained, serverless, zero-config embedded SQL database.
Why: The most deployed database on earth — in every phone, browser, and app.
In-process analytical (OLAP) SQL database — 'SQLite for analytics'.
Why: Redefining local/embedded analytics and data-science workflows.
Open-source Firebase alternative built on Postgres.
Why: Instant backend (auth, storage, realtime, APIs) for indie builders.
Column-oriented database for real-time analytics at scale.
Why: Blazing-fast OLAP powering observability at massive scale.
Self-Hosting & Infrastructure
Own your stack, deploy, orchestrate, and automate on your own servers.
Container platform for packaging and running applications.
Why: The universal unit of deployment and substrate of self-hosting.
Production-grade container orchestration.
Why: The de facto standard for running containers at scale.
Fair-code workflow automation with 400+ integrations and AI nodes.
Why: Self-hosted Zapier alternative — the glue layer for AI automations.
Open-source, self-hostable PaaS (Vercel/Heroku alternative).
Why: Push-to-deploy apps, databases, and 280+ services on your own servers.
Local-first open-source home automation platform.
Why: The privacy-respecting hub of the self-hosted smart home.
Web server with automatic HTTPS by default.
Why: Radically simplifies TLS and reverse-proxy setup.
Open-source infrastructure-as-code (community Terraform fork).
Why: Truly open IaC, Linux Foundation-backed, after Terraform's license change.
Creative & Media
Professional-grade creative tools that happen to be free.
Free and open-source 3D creation suite.
Why: A pro 3D pipeline rivaling paid tools — one of open source's greatest wins.
Feature-packed, fully open-source 2D/3D game engine.
Why: The leading open alternative to Unity/Unreal — no royalties, no traps.
Professional open-source digital painting application.
Why: Artist-built painting tool that rivals commercial software.
Open-source live streaming and screen recording.
Why: The universal standard across the creator economy.
Foundations & Communities
The neutral homes and communities that keep open source alive.
Nonprofit hosting the world's most critical open-source infrastructure.
Why: Neutral home of Linux, Kubernetes, and hundreds of internet-underpinning projects.
Independent nonprofit stewarding the Rust language.
Why: Neutral, sustainable governance behind the safe-systems movement.
Volunteer nonprofit behind 300+ Apache-licensed projects.
Why: Foundational governance model and home to Kafka, Spark, and more.
Home of cloud-native open-source projects.
Why: Vendor-neutral hub for Kubernetes and the cloud-native stack.
Nonprofit supporting the Python language and community.
Why: Stewards the language that powers most of AI/ML and data science.
Platform and community hub for open models, datasets, and ML tooling.
Why: The 'GitHub of AI' — social layer for the whole open-model movement.
Open research collective advancing open models, alignment, and agents.
Why: Creator of the Hermes models and agent — a leading community-driven open-AI org.