AI Tools Radar

About & methodology

AI Tools Radar is a small, non-commercial BranoLabs project. Once a day an agent collects public data about AI repositories and launches and rebuilds this site. Nothing here is written by hand, ranked by payment or endorsed by us.

Sources

SourceItemsFetchedWhat we take
GitHub Trending14021 Oct 2026, 14:13 (Berlin time)Daily and weekly lists for all languages, python, typescript, rust, go, jupyter-notebook. "Stars today / this week" are the numbers GitHub showed at fetch time.
GitHub Search API79621 Oct 2026, 14:15 (Berlin time)Repositories created on or after 2026-09-21: top 100 by stars for each AI topic (llm, ai-agents, ai-agent, agents, agentic-ai, mcp, mcp-server, claude-code, codex, rag, generative-ai, ai, machine-learning, deep-learning, llm-inference, openai, local-llm, agent-skills; >= 20 stars) plus the top 100 new repositories overall (>= 100 stars), then keyword-filtered for AI.
GitHub GraphQL API93621 Oct 2026, 14:17 (Berlin time)Total stars, forks, license, topics, created and last-push dates for every listed repository.
Hacker News Show HN (Algolia HN Search API)8921 Oct 2026, 14:17 (Berlin time)Show HN posts since 2026-09-21 with at least 10 points (287 scanned), keyword-filtered for AI in the title. Points and comment counts as of fetch time.

What counts as an AI tool

A repository is included when its name, description or GitHub topics contain an AI keyword (for example AI, LLM, agent, MCP, RAG, Claude, Codex, inference, diffusion, embeddings). Show HN posts are included when the title contains one. The matched keywords are stored in the JSON data (ai_terms). This is a simple keyword filter: it will miss some AI projects and include some that only mention AI. Forks are excluded.

Categories

Every listed item gets one or more categories, assigned during the daily data build by fixed rules, never by hand and never by an AI model. A rule fires when a repository has a matching GitHub topic (for example mcp-server → MCP, skills & plugins) or when its name or description, or a Show HN title, contains a matching keyword (for example "text-to-speech" → Voice & speech). Jupyter Notebook repositories that match no other rule go to Learning & resources. Items that match no rule are in "Other". The evidence for every label is stored in category_matches in the JSON data. Keyword rules make mistakes: an item that only mentions a word can land in a category it doesn't really belong to. Browse all categories.

Search

The search box on the home page and on category pages searches all lists at once, in your browser: repository names and owners, descriptions, GitHub topics and languages, plus Show HN titles, authors and link domains. Every word you type must match. Filtered views have shareable URLs, for example /?q=agent or /?q=voice&cat=local.

Lists

Caveats

Data & feedback

All data is available as JSON at https://radar.branolabs.com/api/radar.json and summarised for AI assistants in /llms.txt. Last update: 21 Oct 2026, 14:13 (Berlin time). Found a wrong or missing entry? Use "Report a bug" in the footer or reach us at @BranoLabs on X.