Today's GitHub trending list is overwhelmingly shaped by the AI coding-agent ecosystem. Anthropic's financial-services reference repo and the Rust-built DeepSeek-TUI lead on raw star counts, while decolua/9router and addyosmani/agent-skills reflect growing demand for cost-efficient, multi-provider agent setups. Peripheral themes include speculative-decoding research (dflash), stealth browser automation (CloakBrowser), and local LLM-powered research (local-deep-research).
anthropics/financial-services
⭐ 15,799 stars · +3660 today · Python
An Anthropic-published Python repository of reference patterns and code for deploying Claude in financial-services contexts. The blank description suggests early-stage or internal tooling; the high star count likely reflects interest in official Anthropic guidance for a regulated sector.
addyosmani/agent-skills
⭐ 36,010 stars · +1893 today · Shell · topics: agent-skills, antigravity, antigravity-ide, claude-code, cursor, skills
A curated collection of shell-based skill definitions intended to make AI coding agents—Claude Code, Cursor, and others—behave more reliably in production engineering workflows. It targets practitioners tired of agents that work in demos but fail on real codebases.
Hmbown/DeepSeek-TUI
⭐ 22,229 stars · +3731 today · Rust · topics: cli, deepseek, llm, rust, terminal, tui
A terminal user interface written in Rust that lets developers run DeepSeek LLMs as a coding agent directly in the shell without a browser or GUI. The Rust implementation suggests a focus on low overhead and speed.
z-lab/dflash
⭐ 3,910 stars · +379 today · Python
A research implementation of block diffusion combined with speculative decoding, aimed at making large-language-model inference faster. It sits at the intersection of two active inference-speed research threads and likely trended on the back of a paper or preprint release.
decolua/9router
⭐ 5,815 stars · +1052 today · JavaScript · topics: ai-agents, ai-gateway, anthropic, chatgpt, claude, claude-code, cline, codex, copilot, cursor, deepseek, free-ai, gemini, gemini-cli, llm, llm-gateway, openai, openai-proxy, qwen, token-saver
A JavaScript LLM gateway that proxies requests from coding tools like Claude Code, Codex, and Cursor to free-tier API endpoints across 40-plus providers, with automatic failover and built-in token compression. Appeals to developers wanting to cut costs or avoid rate limits.
CloakHQ/CloakBrowser
⭐ 3,322 stars · +526 today · Python · topics: ai-agents, anti-detect, antidetect-browser, bot-detection, browser-automation, captcha-bypass, chromium, cloudflare, cloudflare-bypass, fingerprint, headless-browser, playwright, puppeteer, python, recaptcha, selenium, stealth-browser, undetected, web-scraping, webscraping
A patched Chromium build designed to evade bot-detection systems, offered as a drop-in replacement for Playwright. It claims a clean pass on 30 automated fingerprinting tests, making it relevant for scraping and browser-automation tasks blocked by services like Cloudflare.
awslabs/aidlc-workflows
⭐ 1,820 stars · +58 today · Python
An AWS Labs project that provides adaptive workflow steering rules—essentially structured prompts and process definitions—to guide AI coding agents through software development life-cycle phases. Complements agent IDEs rather than replacing them.
HKUDS/AI-Trader
⭐ 14,823 stars · +202 today · Python
A fully automated trading agent system from HKUDS that uses AI to execute trades end-to-end without human intervention. The 'agent-native' framing suggests it is built around LLM-driven decision loops rather than traditional rule-based algorithms.
LearningCircuit/local-deep-research
⭐ 6,817 stars · +559 today · Python · topics: academia, anthropic, arxiv, brave, deep-research, encryption, home-automation, homeserver, local, local-deep-research, local-llm, mistral, ollama, openai, pubmed, research, research-tool, retrieval-augmented-generation, searxng, self-hosted
A self-hosted research tool that queries over ten search engines—including arXiv and PubMed—and synthesises answers using local or cloud LLMs, with all data encrypted on-device. It benchmarks near 95% on SimpleQA, which likely drew attention.
lobehub/lobehub
⭐ 76,601 stars · +125 today · TypeScript · topics: agent, agent-collaboration, agent-harness, ai, chatgpt, claude, deepseek, gemini, gpt, knowledge-base, mcp, openai
A TypeScript platform for building and coordinating teams of AI agents, supporting multi-agent collaboration and MCP integration across major LLM providers. Its high total star count reflects an established user base; today's uptick is modest relative to the others.
Source: GitHub Trending — Daily [1]