What's Actually
Used in AI Dev

Detects AI coding tools by config files: CLAUDE.md, .cursorrules, AGENTS.md, .windsurfrules, .github/copilot-instructions.md, .aider.conf.yml, .clinerules. Tracks 7 tools: Claude Code, Cursor, OpenAI Codex, GitHub Copilot, Windsurf, Aider, Cline.

1. 1 Claude Code 4,709
2. 2 AGENTS.md (Codex等) 3,261
3. 3 Cursor 340
4. 4 GitHub Copilot 238
5. 5 Cline 23

Primary language breakdown for AI-relevant repos (AI score ≥ 40). Aggregates GitHub's detected `primary_language` to show which languages dominate AI development.

1. 1 Python 9,795
2. 2 TypeScript 4,327
3. 3 JavaScript 3,220
4. 4 HTML 2,085
5. 5 Shell 866

GitHub topic tags self-applied by repo owners. llm/agent/rag/chatbot — what developers tag their AI projects with.

1. 1 claude-code 1,559
2. 2 ai-agents 1,154
3. 3 llm 1,068
4. 4 claude 942
5. 5 python 796

Which LLM provider companies/SDKs are mentioned across observed repos. Counts mentions of OpenAI, Anthropic, Google Gemini, AWS Bedrock, etc.

1. 1 OpenAI 1,124
2. 2 Gemini 1,014
3. 3 Anthropic 686
4. 4 Ollama 384
5. 5 Groq 371

AI-specific frameworks: LangChain, LlamaIndex, AutoGen, CrewAI, etc. (Web/UI frameworks are in a separate category.)

1. 1 LangChain 420
2. 2 LangGraph 379
3. 3 CrewAI 33
4. 4 LlamaIndex 14
5. 5 Instructor 14

General-purpose frameworks for AI app frontends/APIs: Next.js, FastAPI, Streamlit, Gradio, Django, React, Flask, etc.

1. 1 React 815
2. 2 FastAPI 791
3. 3 Streamlit 482
4. 4 Next.js 346
5. 5 Flask 223

Where AI apps/models are deployed: AWS, GCP, Azure, Vercel, Cloudflare, Modal, Replicate, RunPod, Hugging Face Spaces, etc.

1. 1 Vercel 120
2. 2 Cloudflare 108
3. 3 AWS 102
4. 4 Azure 84
5. 5 Railway 40

Authentication layer for AI SaaS: Auth0, Clerk, Supabase Auth, NextAuth, Firebase Auth, WorkOS, etc.

1. 1 Clerk 17
2. 2 Keycloak 6
3. 3 Firebase Auth 3
4. 4 Supabase Auth 2
5. 5 Auth0 2

Vector databases for AI/RAG embeddings storage: Pinecone, Chroma, Qdrant, pgvector, FAISS, etc.

1. 1 Chroma 207
2. 2 pgvector 59
3. 3 Pinecone 36
4. 4 Elasticsearch 8
5. 5 OpenSearch 3

General-purpose databases beyond vector storage: PostgreSQL, MySQL, MongoDB, Redis, Oracle, SQL Server, Supabase, Neon, etc.

1. 1 SQLite 276
2. 2 PostgreSQL 200
3. 3 Supabase 131
4. 4 MongoDB 93
5. 5 Redis 76

Which specific LLM models get mentioned: GPT-4o, Claude 3.7, Gemini 2, Llama 3.x, DeepSeek-V3, etc.

1. 1 Claude 6,380
2. 2 Gemini 1,014
3. 3 Llama 608
4. 4 DeepSeek 329
5. 5 Qwen 181

Embedding model adoption: text-embedding-3, nomic-embed, bge, jina, etc.

1. 1 sentence-transformers 22
2. 2 bge 12
3. 3 voyage 9
4. 4 gte 4
5. 5 text-embedding-3 3

AI agent construction frameworks: LangGraph, CrewAI, AutoGen, smolagents, MCP, etc.

1. 1 MCP 1,510
2. 2 AgentOps 11
3. 3 AutoGPT 2
4. 4 phidata 1