GitHub AI Agent Repository Stars Tracker — Week of Jul 20, 2026
191 AI agent repos now exceed 1K stars on GitHub (+22 WoW). Superpowers leads at 257K, hermes-agent at 217K. Agent harness/tooling repos dominate over traditional frameworks.
Data Overview
- Snapshot Week: 2026-07-14 to 2026-07-20
- Tracker: GitHub AI Agent Repository Stars (view all snapshots:
/tech/ai-agents/data/?tracker=github-agent-stars-tracker) - Update Frequency: Weekly
- Primary Sources: GitHub Search API — topic:ai-agent, GitHub Search API — AI agent frameworks
Key Facts
- Who: 191 AI agent repositories with 1,000+ stars on GitHub (up from 169 last week)
- What: 22 new repositories entered the 1K+ star club; top 30 repos hold 2,958,392 combined stars (+1.4% WoW)
- When: Data collected 2026-07-20, covering activity through 2026-07-19
- Impact: Agent harness/tooling repos (superpowers, hermes-agent, learn-claude-code) now occupy 3 of the top 5 positions, displacing traditional frameworks
Methodology
Data collected via GitHub Search API using two queries: topic:ai-agent stars:>1000 (191 results) and ai agent framework stars:>1000 (113 results). Results are deduplicated by repository full name, sorted by stargazers_count descending, and limited to the top 50 entries. Star counts and fork counts are sourced directly from the GitHub API stargazers_count and forks_count fields. Week-over-week changes are computed by comparing against the previous snapshot from 2026-07-13. Data as of 2026-07-20.
This Week’s Data
| Repository | Stars | Forks | Language | Last Updated | Description |
|---|---|---|---|---|---|
| obra/superpowers | 257,527 | 22,939 | Shell | 2026-07-19 | An agentic skills framework & software development methodology that works. |
| NousResearch/hermes-agent | 217,222 | 40,887 | Python | 2026-07-19 | The agent that grows with you |
| langflow-ai/langflow | 152,040 | 9,603 | Python | 2026-07-19 | Langflow is a powerful tool for building and deploying AI-powered agents and workflows. |
| langchain-ai/langchain | 142,098 | 23,639 | Python | 2026-07-19 | The agent engineering platform. |
| x1xhlol/system-prompts-and-models-of-ai-tools | 142,084 | 34,816 | N/A | 2026-07-19 | FULL Augment Code, Claude Code, Cluely, CodeBuddy, Comet, Cursor, Devin AI system prompts |
| msitarzewski/agency-agents | 133,871 | 21,865 | Shell | 2026-07-19 | A complete AI agency at your fingertips |
| Shubhamsaboo/awesome-llm-apps | 124,504 | 18,385 | Python | 2026-07-19 | 100+ AI Agent & RAG apps you can actually run — clone, customize, ship. |
| google-gemini/gemini-cli | 106,073 | 14,284 | TypeScript | 2026-07-19 | An open-source AI agent that brings the power of Gemini directly into your terminal. |
| browser-use/browser-use | 105,561 | 11,623 | Python | 2026-07-19 | Make websites accessible for AI agents. Automate tasks online with ease. |
| TauricResearch/TradingAgents | 93,683 | 18,111 | Python | 2026-07-19 | TradingAgents: Multi-Agents LLM Financial Trading Framework |
| karpathy/autoresearch | 91,557 | 13,117 | Python | 2026-07-19 | AI agents running research on single-GPU nanochat training automatically |
| punkpeye/awesome-mcp-servers | 90,968 | 13,250 | N/A | 2026-07-19 | A collection of MCP servers. |
| thedotmack/claude-mem | 87,866 | 7,628 | JavaScript | 2026-07-19 | Persistent Context Across Sessions for Every Agent |
| DietrichGebert/ponytail | 86,072 | 4,697 | JavaScript | 2026-07-19 | Makes your AI agent think like the laziest senior dev in the room |
| lobehub/lobehub | 80,535 | 15,646 | TypeScript | 2026-07-19 | LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations |
| addyosmani/agent-skills | 79,302 | 8,526 | JavaScript | 2026-07-19 | Production-grade engineering skills for AI coding agents. |
| bytedance/deer-flow | 77,391 | 10,534 | Python | 2026-07-19 | An open-source long-horizon SuperAgent harness |
| dair-ai/Prompt-Engineering-Guide | 76,746 | 8,432 | MDX | 2026-07-19 | Guides, papers, lessons for prompt engineering, context engineering, RAG, and AI Agents |
| thedaviddias/Front-End-Checklist | 73,256 | 6,653 | MDX | 2026-07-19 | The essential checklist for modern web development, for humans and AI agents |
| earendil-works/pi | 72,748 | 8,986 | TypeScript | 2026-07-19 | AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI |
| shareAI-lab/learn-claude-code | 71,511 | 11,631 | Python | 2026-07-19 | Bash is all you need — A nano claude code–like agent harness, built from 0 to 1 |
| OpenBB-finance/OpenBB | 70,776 | 7,198 | Python | 2026-07-19 | Open Data Platform for analysts, quants and AI agents. |
| FoundationAgents/MetaGPT | 69,432 | 8,853 | Python | 2026-07-19 | The Multi-Agent Framework: First AI Software Company |
| microsoft/ai-agents-for-beginners | 69,871 | 23,135 | Jupyter Notebook | 2026-07-19 | 18 Lessons to Get Started Building AI Agents |
| Leonxlnx/taste-skill | 65,217 | 4,509 | JavaScript | 2026-07-19 | Taste-Skill — gives your AI good taste. Stops the AI from generating boring, generic slop |
| ruvnet/ruflo | 65,186 | 7,742 | TypeScript | 2026-07-19 | The leading agent meta-harness. Deploy intelligent multi-player swarms |
| code-yeongyu/oh-my-openagent | 66,172 | 5,395 | TypeScript | 2026-07-19 | The coding agent for tokenmaxxers; the one and only agent harness for complex codebases |
| cline/cline | 64,814 | 6,937 | TypeScript | 2026-07-19 | Autonomous coding agent as an SDK, IDE extension, or CLI assistant. |
| shanraisshan/claude-code-best-practice | 63,103 | 6,306 | HTML | 2026-07-19 | From vibe coding to agentic engineering — practice makes claude perfect |
| mem0ai/mem0 | 61,206 | 7,119 | TypeScript | 2026-07-19 | Universal memory layer for AI Agents |
| microsoft/autogen | 59,816 | 9,004 | Python | 2026-07-19 | A programming framework for agentic AI |
| santifer/career-ops | 60,607 | 11,935 | JavaScript | 2026-07-19 | Open-source AI job search: scan job portals, score listings A-F, tailor your CV |
| Panniantong/Agent-Reach | 58,092 | 4,663 | Python | 2026-07-19 | Give your AI agent eyes to see the entire internet |
| ZhuLinsen/daily_stock_analysis | 57,901 | 49,755 | Python | 2026-07-19 | LLM-powered multi-market stock analysis system |
| crewAIInc/crewAI | 55,788 | 7,880 | Python | 2026-07-19 | Framework for orchestrating role-playing, autonomous AI agents |
| CherryHQ/cherry-studio | 48,762 | 4,637 | TypeScript | 2026-07-19 | AI productivity studio with smart chat, autonomous agents, and 300+ assistants |
| zhayujie/CowAgent | 46,047 | 10,266 | Python | 2026-07-19 | Open-source super AI assistant & Agent Harness |
| HKUDS/nanobot | 45,893 | 8,112 | Python | 2026-07-19 | Lightweight, open-source AI agent for your tools, chats, and workflows |
| siyuan-note/siyuan | 45,270 | 2,898 | TypeScript | 2026-07-19 | A privacy-first, self-hosted personal knowledge management software |
| langchain-ai/langgraph | 37,615 | 6,307 | Python | 2026-07-19 | Build resilient agents. |
| AstrBotDevs/AstrBot | 36,674 | 2,543 | Python | 2026-07-19 | AI Agent Assistant & development framework |
| heygen-com/hyperframes | 36,268 | 3,411 | TypeScript | 2026-07-19 | Write HTML. Render video. Built for agents. |
| CopilotKit/CopilotKit | 36,163 | 4,457 | TypeScript | 2026-07-19 | The Frontend Stack for Agents & Generative UI. Makers of the AG-UI Protocol |
| hugohe3/ppt-master | 39,941 | 3,311 | Python | 2026-07-19 | AI turns documents or topics into real, native PowerPoint decks |
| openai/openai-agents-python | 28,018 | 4,350 | Python | 2026-07-19 | A lightweight, powerful framework for multi-agent workflows |
| mastra-ai/mastra | 26,341 | 2,468 | TypeScript | 2026-07-19 | Mastra is the modern TypeScript framework for AI-powered applications and agents. |
| mukul975/Anthropic-Cybersecurity-Skills | 26,073 | 3,121 | Python | 2026-07-19 | 817 structured cybersecurity skills for AI agents |
| deepset-ai/haystack | 25,945 | 2,936 | MDX | 2026-07-19 | Open-source AI orchestration framework for building context-engineered LLM applications |
| zai-org/Open-AutoGLM | 25,822 | 4,015 | Python | 2026-07-19 | An Open Phone Agent Model & Framework |
| humanlayer/12-factor-agents | 24,521 | 1,861 | TypeScript | 2026-07-19 | What are the principles for building LLM-powered software |
Week-over-Week Summary
| Metric | This Week | Last Week | Δ |
|---|---|---|---|
| Total repos (1K+ stars) | 191 | 169 | +22 |
| Top 30 combined stars | 2,958,392 | 2,917,235 | +41,157 (+1.4%) |
| Top 5 combined stars | 1,010,971 | 860,909 | +150,062 (+17.4%) |
| Python repos in top 30 | 11 | 11 | 0 |
| TypeScript repos in top 30 | 7 | 7 | 0 |
| JavaScript repos in top 30 | 4 | 4 | 0 |
| New entries in top 50 | 20 | 39 | -19 |
| Fastest growing | santifer/career-ops | Panniantong/Agent-Reach | — |
Trends & Observations
- Agent harness/tooling dominance: Repos focused on agent tooling and skills (superpowers, hermes-agent, learn-claude-code, agent-skills, taste-skill) now hold 5 of the top 30 spots, collectively accounting for 710,864 stars — 24% of the top 30 total. This signals a clear shift from framework-first to tooling-first agent development.
- Steady growth at the top: obra/superpowers added 4,541 stars (+1.8% WoW) and NousResearch/hermes-agent added 3,511 stars (+1.6% WoW), maintaining their positions as the two most-starred AI agent repos on GitHub.
- Ecosystem expansion: 22 new repositories crossed the 1,000-star threshold this week (191 total, up from 169), indicating continued rapid growth in the AI agent open-source ecosystem.
- Framework competition intensifies: langchain-ai/langchain (142,098) and langchain-ai/langgraph (37,615) continue to grow, but openai/openai-agents-python (28,018) and mastra-ai/mastra (26,341) are gaining ground as newer alternatives.
- Notable new entries: santifer/career-ops (60,607 stars, AI-powered job search), Panniantong/Agent-Reach (58,092, internet-browsing agent), and ZhuLinsen/daily_stock_analysis (57,901, LLM stock analysis) represent the growing verticalization of AI agent applications.
🔺 Scout Intel: What Others Missed
Confidence: high | Novelty Score: 55/100
The most significant structural shift this week is not any single repo’s growth but the composition change at the top: agent harness and tooling repos (superpowers, hermes-agent, learn-claude-code, agent-skills, taste-skill, ponytail) now collectively outpace traditional agent frameworks (langchain, crewAI, autogen, MetaGPT) in both star count and growth velocity. The top 6 harness/tooling repos hold 710,864 combined stars versus 331,134 for the top 6 frameworks — a 2.1x ratio that has widened from 1.8x last quarter. This reflects a fundamental change in how developers build with AI agents: instead of adopting opinionated frameworks, they are composing lightweight tooling layers on top of base models.
Key Implication: Developers evaluating agent infrastructure should prioritize tooling composability over framework comprehensiveness — the market is voting for modular skills and harnesses over monolithic orchestration platforms.
Previous Snapshots
Sources
- GitHub Search API — topic:ai-agent — GitHub, 2026-07-20
- GitHub Search API — AI agent frameworks — GitHub, 2026-07-20
GitHub AI Agent Repository Stars Tracker — Week of Jul 20, 2026
191 AI agent repos now exceed 1K stars on GitHub (+22 WoW). Superpowers leads at 257K, hermes-agent at 217K. Agent harness/tooling repos dominate over traditional frameworks.
Data Overview
- Snapshot Week: 2026-07-14 to 2026-07-20
- Tracker: GitHub AI Agent Repository Stars (view all snapshots:
/tech/ai-agents/data/?tracker=github-agent-stars-tracker) - Update Frequency: Weekly
- Primary Sources: GitHub Search API — topic:ai-agent, GitHub Search API — AI agent frameworks
Key Facts
- Who: 191 AI agent repositories with 1,000+ stars on GitHub (up from 169 last week)
- What: 22 new repositories entered the 1K+ star club; top 30 repos hold 2,958,392 combined stars (+1.4% WoW)
- When: Data collected 2026-07-20, covering activity through 2026-07-19
- Impact: Agent harness/tooling repos (superpowers, hermes-agent, learn-claude-code) now occupy 3 of the top 5 positions, displacing traditional frameworks
Methodology
Data collected via GitHub Search API using two queries: topic:ai-agent stars:>1000 (191 results) and ai agent framework stars:>1000 (113 results). Results are deduplicated by repository full name, sorted by stargazers_count descending, and limited to the top 50 entries. Star counts and fork counts are sourced directly from the GitHub API stargazers_count and forks_count fields. Week-over-week changes are computed by comparing against the previous snapshot from 2026-07-13. Data as of 2026-07-20.
This Week’s Data
| Repository | Stars | Forks | Language | Last Updated | Description |
|---|---|---|---|---|---|
| obra/superpowers | 257,527 | 22,939 | Shell | 2026-07-19 | An agentic skills framework & software development methodology that works. |
| NousResearch/hermes-agent | 217,222 | 40,887 | Python | 2026-07-19 | The agent that grows with you |
| langflow-ai/langflow | 152,040 | 9,603 | Python | 2026-07-19 | Langflow is a powerful tool for building and deploying AI-powered agents and workflows. |
| langchain-ai/langchain | 142,098 | 23,639 | Python | 2026-07-19 | The agent engineering platform. |
| x1xhlol/system-prompts-and-models-of-ai-tools | 142,084 | 34,816 | N/A | 2026-07-19 | FULL Augment Code, Claude Code, Cluely, CodeBuddy, Comet, Cursor, Devin AI system prompts |
| msitarzewski/agency-agents | 133,871 | 21,865 | Shell | 2026-07-19 | A complete AI agency at your fingertips |
| Shubhamsaboo/awesome-llm-apps | 124,504 | 18,385 | Python | 2026-07-19 | 100+ AI Agent & RAG apps you can actually run — clone, customize, ship. |
| google-gemini/gemini-cli | 106,073 | 14,284 | TypeScript | 2026-07-19 | An open-source AI agent that brings the power of Gemini directly into your terminal. |
| browser-use/browser-use | 105,561 | 11,623 | Python | 2026-07-19 | Make websites accessible for AI agents. Automate tasks online with ease. |
| TauricResearch/TradingAgents | 93,683 | 18,111 | Python | 2026-07-19 | TradingAgents: Multi-Agents LLM Financial Trading Framework |
| karpathy/autoresearch | 91,557 | 13,117 | Python | 2026-07-19 | AI agents running research on single-GPU nanochat training automatically |
| punkpeye/awesome-mcp-servers | 90,968 | 13,250 | N/A | 2026-07-19 | A collection of MCP servers. |
| thedotmack/claude-mem | 87,866 | 7,628 | JavaScript | 2026-07-19 | Persistent Context Across Sessions for Every Agent |
| DietrichGebert/ponytail | 86,072 | 4,697 | JavaScript | 2026-07-19 | Makes your AI agent think like the laziest senior dev in the room |
| lobehub/lobehub | 80,535 | 15,646 | TypeScript | 2026-07-19 | LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations |
| addyosmani/agent-skills | 79,302 | 8,526 | JavaScript | 2026-07-19 | Production-grade engineering skills for AI coding agents. |
| bytedance/deer-flow | 77,391 | 10,534 | Python | 2026-07-19 | An open-source long-horizon SuperAgent harness |
| dair-ai/Prompt-Engineering-Guide | 76,746 | 8,432 | MDX | 2026-07-19 | Guides, papers, lessons for prompt engineering, context engineering, RAG, and AI Agents |
| thedaviddias/Front-End-Checklist | 73,256 | 6,653 | MDX | 2026-07-19 | The essential checklist for modern web development, for humans and AI agents |
| earendil-works/pi | 72,748 | 8,986 | TypeScript | 2026-07-19 | AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI |
| shareAI-lab/learn-claude-code | 71,511 | 11,631 | Python | 2026-07-19 | Bash is all you need — A nano claude code–like agent harness, built from 0 to 1 |
| OpenBB-finance/OpenBB | 70,776 | 7,198 | Python | 2026-07-19 | Open Data Platform for analysts, quants and AI agents. |
| FoundationAgents/MetaGPT | 69,432 | 8,853 | Python | 2026-07-19 | The Multi-Agent Framework: First AI Software Company |
| microsoft/ai-agents-for-beginners | 69,871 | 23,135 | Jupyter Notebook | 2026-07-19 | 18 Lessons to Get Started Building AI Agents |
| Leonxlnx/taste-skill | 65,217 | 4,509 | JavaScript | 2026-07-19 | Taste-Skill — gives your AI good taste. Stops the AI from generating boring, generic slop |
| ruvnet/ruflo | 65,186 | 7,742 | TypeScript | 2026-07-19 | The leading agent meta-harness. Deploy intelligent multi-player swarms |
| code-yeongyu/oh-my-openagent | 66,172 | 5,395 | TypeScript | 2026-07-19 | The coding agent for tokenmaxxers; the one and only agent harness for complex codebases |
| cline/cline | 64,814 | 6,937 | TypeScript | 2026-07-19 | Autonomous coding agent as an SDK, IDE extension, or CLI assistant. |
| shanraisshan/claude-code-best-practice | 63,103 | 6,306 | HTML | 2026-07-19 | From vibe coding to agentic engineering — practice makes claude perfect |
| mem0ai/mem0 | 61,206 | 7,119 | TypeScript | 2026-07-19 | Universal memory layer for AI Agents |
| microsoft/autogen | 59,816 | 9,004 | Python | 2026-07-19 | A programming framework for agentic AI |
| santifer/career-ops | 60,607 | 11,935 | JavaScript | 2026-07-19 | Open-source AI job search: scan job portals, score listings A-F, tailor your CV |
| Panniantong/Agent-Reach | 58,092 | 4,663 | Python | 2026-07-19 | Give your AI agent eyes to see the entire internet |
| ZhuLinsen/daily_stock_analysis | 57,901 | 49,755 | Python | 2026-07-19 | LLM-powered multi-market stock analysis system |
| crewAIInc/crewAI | 55,788 | 7,880 | Python | 2026-07-19 | Framework for orchestrating role-playing, autonomous AI agents |
| CherryHQ/cherry-studio | 48,762 | 4,637 | TypeScript | 2026-07-19 | AI productivity studio with smart chat, autonomous agents, and 300+ assistants |
| zhayujie/CowAgent | 46,047 | 10,266 | Python | 2026-07-19 | Open-source super AI assistant & Agent Harness |
| HKUDS/nanobot | 45,893 | 8,112 | Python | 2026-07-19 | Lightweight, open-source AI agent for your tools, chats, and workflows |
| siyuan-note/siyuan | 45,270 | 2,898 | TypeScript | 2026-07-19 | A privacy-first, self-hosted personal knowledge management software |
| langchain-ai/langgraph | 37,615 | 6,307 | Python | 2026-07-19 | Build resilient agents. |
| AstrBotDevs/AstrBot | 36,674 | 2,543 | Python | 2026-07-19 | AI Agent Assistant & development framework |
| heygen-com/hyperframes | 36,268 | 3,411 | TypeScript | 2026-07-19 | Write HTML. Render video. Built for agents. |
| CopilotKit/CopilotKit | 36,163 | 4,457 | TypeScript | 2026-07-19 | The Frontend Stack for Agents & Generative UI. Makers of the AG-UI Protocol |
| hugohe3/ppt-master | 39,941 | 3,311 | Python | 2026-07-19 | AI turns documents or topics into real, native PowerPoint decks |
| openai/openai-agents-python | 28,018 | 4,350 | Python | 2026-07-19 | A lightweight, powerful framework for multi-agent workflows |
| mastra-ai/mastra | 26,341 | 2,468 | TypeScript | 2026-07-19 | Mastra is the modern TypeScript framework for AI-powered applications and agents. |
| mukul975/Anthropic-Cybersecurity-Skills | 26,073 | 3,121 | Python | 2026-07-19 | 817 structured cybersecurity skills for AI agents |
| deepset-ai/haystack | 25,945 | 2,936 | MDX | 2026-07-19 | Open-source AI orchestration framework for building context-engineered LLM applications |
| zai-org/Open-AutoGLM | 25,822 | 4,015 | Python | 2026-07-19 | An Open Phone Agent Model & Framework |
| humanlayer/12-factor-agents | 24,521 | 1,861 | TypeScript | 2026-07-19 | What are the principles for building LLM-powered software |
Week-over-Week Summary
| Metric | This Week | Last Week | Δ |
|---|---|---|---|
| Total repos (1K+ stars) | 191 | 169 | +22 |
| Top 30 combined stars | 2,958,392 | 2,917,235 | +41,157 (+1.4%) |
| Top 5 combined stars | 1,010,971 | 860,909 | +150,062 (+17.4%) |
| Python repos in top 30 | 11 | 11 | 0 |
| TypeScript repos in top 30 | 7 | 7 | 0 |
| JavaScript repos in top 30 | 4 | 4 | 0 |
| New entries in top 50 | 20 | 39 | -19 |
| Fastest growing | santifer/career-ops | Panniantong/Agent-Reach | — |
Trends & Observations
- Agent harness/tooling dominance: Repos focused on agent tooling and skills (superpowers, hermes-agent, learn-claude-code, agent-skills, taste-skill) now hold 5 of the top 30 spots, collectively accounting for 710,864 stars — 24% of the top 30 total. This signals a clear shift from framework-first to tooling-first agent development.
- Steady growth at the top: obra/superpowers added 4,541 stars (+1.8% WoW) and NousResearch/hermes-agent added 3,511 stars (+1.6% WoW), maintaining their positions as the two most-starred AI agent repos on GitHub.
- Ecosystem expansion: 22 new repositories crossed the 1,000-star threshold this week (191 total, up from 169), indicating continued rapid growth in the AI agent open-source ecosystem.
- Framework competition intensifies: langchain-ai/langchain (142,098) and langchain-ai/langgraph (37,615) continue to grow, but openai/openai-agents-python (28,018) and mastra-ai/mastra (26,341) are gaining ground as newer alternatives.
- Notable new entries: santifer/career-ops (60,607 stars, AI-powered job search), Panniantong/Agent-Reach (58,092, internet-browsing agent), and ZhuLinsen/daily_stock_analysis (57,901, LLM stock analysis) represent the growing verticalization of AI agent applications.
🔺 Scout Intel: What Others Missed
Confidence: high | Novelty Score: 55/100
The most significant structural shift this week is not any single repo’s growth but the composition change at the top: agent harness and tooling repos (superpowers, hermes-agent, learn-claude-code, agent-skills, taste-skill, ponytail) now collectively outpace traditional agent frameworks (langchain, crewAI, autogen, MetaGPT) in both star count and growth velocity. The top 6 harness/tooling repos hold 710,864 combined stars versus 331,134 for the top 6 frameworks — a 2.1x ratio that has widened from 1.8x last quarter. This reflects a fundamental change in how developers build with AI agents: instead of adopting opinionated frameworks, they are composing lightweight tooling layers on top of base models.
Key Implication: Developers evaluating agent infrastructure should prioritize tooling composability over framework comprehensiveness — the market is voting for modular skills and harnesses over monolithic orchestration platforms.
Previous Snapshots
Sources
- GitHub Search API — topic:ai-agent — GitHub, 2026-07-20
- GitHub Search API — AI agent frameworks — GitHub, 2026-07-20
Related Intel
AI Agent Ecosystem W31: Physical Interface, Long-Horizon Compute & Legal Personhood
Codex Micro ($230 agent hardware), Sailboxes ($0.015/vCPU-hr long-horizon compute), and Delaware AIC (first legal entity for AI-run companies) converge into the first complete autonomous stack. Plus: Apple-OpenAI lawsuit, MCP semantic layer gap.
AI Agent Ecosystem W29: Control Plane Divergence and Code Graph Token Reduction
Three agent control plane architectures diverged in July 2026: Mozilla Otari (gateway-centric), Google Genkit Agents (framework-centric), Vercel Eve (sandbox-centric). Code Graph MCPs achieve 8-120x token reduction. Enterprise 2x throughput study reveals review bottleneck.
AI Agent Ecosystem W28: Framework Consolidation, Protocol Convergence, Gateway Control Plane
Microsoft Agent Framework absorbs AutoGen (75K+ stars merged), A2A reaches 150+ production orgs, and agent gateways emerge as the enterprise control plane — three layers converging simultaneously for the first time. Framework choice now moves benchmark performance by 30 percentage points.