📈 GitHub Trending — August 8, 2026

Daily and weekly trending repos ranked by stars and growth rate. Data sourced from GitHub trending page.

🔥 Daily Trending — August 8, 2026

#	Repository	⭐ Stars	Daily Growth	Category
1	PrimeIntellect-ai/prime-agent VIRAL	6,450	+2,293/day	🧠 RLM Agent Framework (NEW)
2	addyosmani/agent-skills	83,882	+1,131/day	🧩 Agent Skills Framework
3	cloudflare/computer	5,678	+872/day	💻 Agent Hardware Access
4	semantica-agi/semantica NEW	2,340	+122/day	📊 Graph-Native AI Infrastructure (NEW)
5	666ghj/MiroFish NEW	~5,000	+500+/day	🐟 Swarm Intelligence AI (NEW)
6	pranshuparmar/witr	19,727	+234/day	🔍 Process Tracing (ZERO-COMP)
7	denoland/celld	2,196	+516/day	🏠 Durable Objects
8	unclebob/swarm-forge NEW	1,827	+81/day	🐝 Swarm Agent Coordination (NEW)
9	K2SOsint/Legendary_OSINT	1,421	+109/day	🔍 OSINT Tools
10	chenyme/grok2api	~800	Growing	🔌 Grok API Gateway
11	google/skills NEW	New	Launching	🧩 Google .skills Format (NEW)

📅 Weekly Trending — Last 7 Days

#	Repository	⭐ Stars	Weekly Growth	Category
1	zhaoxuya520/reverse-skill VIRAL	20,435	+10,091/week	🔥 Reverse Engineering Skill Router
2	microsoft/AI-For-Beginners	63,104	+9,164/week	📚 AI Education
3	TencentCloud/TencentDB-Agent-Memory	17,388	+6,444/week	🧠 Team-Level AI Memory
4	block/buzz	24,810	+5,903/week	💬 Hive Mind Communication
5	lyogavin/airllm	29,863	+5,222/week	🖥️ Local AI Inference
6	esengine/DeepSeek-Reasonix	32,865	+4,203/week	🖥️ DeepSeek-Native Terminal Agent
7	virgiliojr94/book-to-skill	18,170	+3,903/week	📖 Book-to-Skill
8	ayghri/i-have-adhd	18,070	+3,628/week	🧠 ADHD-Friendly Agent Output
9	usekaneo/kaneo HOT	7,618	+2,925/week	📋 Project Management
10	different-ai/openwork	21,426	+2,939/week	🤖 Open-Source Claude Cowork
11	paperswithbacktest/awesome-systematic-trading NEW	12,939	+1,433/week	📊 Systematic Trading
12	antirez/ds4	20,874	+1,319/week	🖥️ DeepSeek Local Inference
13	DataExpert-io/data-engineer-handbook	43,359	+716/week	📚 Data Engineering
14	tailwindlabs/tailwindcss	97,071	+992/week	🎨 CSS Framework
15	embabel/embabel-agent	4,010	+154/week	☕ JVM Agent Framework

🆕 Fresh Repos — Zero-Competition Opportunities

Repository	⭐	Category	Opportunity
PrimeIntellect-ai/prime-agent	6,450	🧠 RLM Agent Framework (NEW)	🚀 ZERO-COMP — Self-improving RLM agent
google/skills	New	🧩 Google .skills Format (NEW)	🚀 ZERO-COMP — Big Tech's official agent-skills spec
666ghj/MiroFish	~5,000	🐟 Swarm Intelligence (NEW)	🚀 ZERO-COMP — Universal prediction engine
semantica-agi/semantica	2,340	📊 Graph-Native AI (NEW)	🚀 ZERO-COMP — Deterministic context graphs
pranshuparmar/witr	19,727	🔍 Process Tracing (ZERO-COMP)	🚀 ZERO-COMP — "Why is this running?"
unclebob/swarm-forge	1,827	🐝 Swarm Coordination (NEW)	🚀 ZERO-COMP — tmux-based agent orchestration
denoland/celld	2,196	🏠 Durable Objects	📈 Growing — Self-hosted distributed DOs
chenyme/grok2api	~800	🔌 Grok API Gateway	📈 Growing — Multi-account Grok API
K2SOsint/Legendary_OSINT	1,421	🔍 OSINT Tools	📈 Growing — Curated OSINT stack

🔑 Key Trends — August 8, 2026

🧠 RLM (Recursive Language Model) cements itself as the #1 new category: PrimeAgent (6,450⭐, +2,293/day) crossed the 6K mark in under two weeks. Context-as-variables + tools-as-functions is a paradigm shift with zero comparison content on the web.

🏛️ Google enters the agent-skills war — google/skills just launched: After addyosmani/agent-skills (83,882⭐) and the .agents spec, Google shipping its own skills format legitimizes the category. The format war is now officially three-way: Anthropic .agents, Google .skills, community .md.

🐟 Swarm Intelligence for AI Prediction keeps compounding: MiroFish (~5,000⭐) remains the category leader at "predict anything" scale. unclebob/swarm-forge (1,827⭐, +81/day) adds disciplined orchestration. Zero competition for cross-comparison content.

🏠 Durable Objects is the quiet infrastructure winner: denoland/celld (2,196⭐, +516/day) is compounding hard — self-hosted distributed durable objects are the new serverless primitive. Still under-covered territory.

📊 Systematic trading crosses 12.9K: awesome-systematic-trading (12,939⭐, +1,433/week) keeps climbing — quant/agent crossover content remains an open lane.

🔍 Process Tracing keeps its zero-competition moat: witr (19,727⭐, +234/day) — "Why is this running?" in one command. No serious comparison content exists anywhere.