Preview mode — 111 repos, zero database

The open-source stack,
curated by builders.

Tell us what you're building or what you need. We'll surface the right repo, with an honest take on whether to use it.

Filtered byAI & MLclear

AI & ML

12 repos

The biggest LLM framework. Chains, agents, RAG pipelines, integrations with every model, vector DB, and tool. Strong ecosystem with LangGraph for agentic workflows. The criticism: for simple use cases…

AI & ML

You are building multi-step AI pipelines (agents, complex RAG, tool use) and need the ecosystem.

You are just calling an LLM API and processing the response — use the native SDK directly.

langchain-ai/langchainView

run-llama

llama_index

RAG specialist. Better document loading, chunking, and retrieval than LangChain for retrieval-first apps. Strong on advanced strategies — hybrid search, re-ranking, recursive retrieval. Now expanded i…

AI & ML

Your AI app is mostly retrieval over documents — knowledge base, support bot, internal search.

You need broad agent/tool use — LangChain has wider integration coverage.

run-llama/llama_indexView

microsoft

autogen

Multi-agent conversation framework from Microsoft. Stronger than LangChain for genuinely autonomous multi-agent setups where agents negotiate with each other. Younger ecosystem, fewer integrations.

AI & ML

You are building experimental multi-agent systems where agents collaborate or debate.

You just want a single agent or a simple chatbot — overkill.

microsoft/autogenView

crewAIInc

crewAI

Role-based agent orchestration. Simpler mental model than AutoGen — define agents with roles, give them tools, let them collaborate. Popular for content workflows, research agents, and internal automa…

AI & ML

You want multi-agent collaboration with a simple role-based mental model.

You need fine-grained control over agent communication or large-scale production deployment.

crewAIInc/crewAIView

qdrant

Rust-based vector database — fast, production-grade, Apache 2.0. Good developer experience and clear pricing if you use their managed cloud. Strong choice if Postgres+pgvector cannot handle your scale…

AI & MLDatabase & ORM

You need production-grade vector search at scale and Postgres is hitting limits.

You are starting fresh on Postgres — pgvector is enough for most apps.

qdrant/qdrantView

chroma-core

chroma

AI-native embedding database designed for rapid prototyping. Easy to run locally, tight LangChain integration. Better for hackathons and internal tools than huge production deployments.

AI & MLDatabase & ORM

You are prototyping RAG, building an internal tool, or bundling a vector DB into a desktop/local app.

You expect millions of vectors in production — look at Qdrant, Weaviate, or pgvector.

chroma-core/chromaView

pgvector

Vector search inside Postgres. ~12k stars and dominant in 2026 as the boring-but-correct default. If you already have Postgres (Supabase, Neon, RDS), do not add another database — just install this ex…

AI & MLDatabase & ORM

You are already on Postgres and want vector search without operating a second database.

You need billion-scale vector search with millisecond latency — a dedicated vector DB is worth the operational cost.

pgvector/pgvectorView

milvus-io

milvus

Open-source vector DB with the most mature features for huge scale. ~30k stars. Steeper operational burden than Qdrant. Used in production by companies doing serious vector workloads.

AI & MLDatabase & ORM

You are running vector search at extreme scale with a team that can operate distributed systems.

You are a small team — Qdrant or pgvector will be enough and easier to run.

milvus-io/milvusView

weaviate

Vector DB with strong hybrid search (vector + keyword) and built-in modules for common embedding workflows. Slightly higher learning curve than Qdrant but powerful for production RAG.

AI & MLDatabase & ORM

You need hybrid keyword + vector search and want it built in rather than bolted on.

You only need pure vector search — Qdrant is simpler.

weaviate/weaviateView

ollama

Run open-source LLMs locally with one command. ~95k stars. Essential for AI development — test prompts against Llama, Mistral, Phi without burning API credits. Production use is limited unless you hav…

AI & MLCLI & Dev Tools

You want to develop and test LLM features without API costs, or run small models on a workstation.

You need production inference at scale — use vLLM or a hosted API.

ollama/ollamaView

mendableai

firecrawl

The 2026 default for LLM-grade scraping. Fires Playwright behind the scenes, returns clean markdown ready to feed into RAG. Self-host for free or use the hosted tier — both expose the same API. The sw…

Scraping & CrawlingAI & ML

You're building a RAG agent, a competitive-intel tool, or any system that ingests web content as training/search data.

You're scraping structured APIs (JSON endpoints) — overkill. Use plain fetch + Zod.

mendableai/firecrawlView

unclecode

crawl4ai

Python-native LLM-friendly crawler. Strong at extracting structured data (JSON schemas) from messy HTML using an embedded LLM. Heavier setup than Firecrawl but more control over extraction prompts. Be…

Scraping & CrawlingAI & ML

You're in Python, want self-hosted, and need extraction with an exact JSON schema.

You don't need LLM-driven extraction — Scrapy or Crawlee will be cheaper and faster.

unclecode/crawl4aiView

Don't see what you need?

We'll add it in 60 minutes.

Tell us what tool or use case is missing. We'll research the best repo for it, write an honest take, add it to the directory, and email you the link. No paywall, no signup required.

The open-source stack,curated by builders.

AI & ML

We'll add it in 60 minutes.

The open-source stack,
curated by builders.