Data Overview

• Snapshot Week: 2026-05-04 to 2026-05-11
• Tracker: GitHub AI Agent Stars Tracker (view all historical snapshots: /tech/ai-agents/data/?tracker=github-agent-stars-tracker)
• Update Frequency: Weekly
• Primary Sources: GitHub Search API — ai-agent topic, stars > 1000

Key Facts

• Who: 30 AI agent repositories tracked via GitHub’s ai-agent topic search
• What: Dramatic leaderboard reshuffle — Hermes Agent (NousResearch) emerged as new #1 at 142,366 stars; previous top 5 all disappeared from topic search results
• When: Data collected 2026-05-11, reflecting repository states as of May 10, 2026
• Impact: Complete transformation of the ai-agent ecosystem visibility landscape; Claude Code-compatible frameworks now dominate the top positions

Methodology

Data Collection: Repositories retrieved via GitHub Search API using the topic:ai-agent filter with a minimum threshold of 1,000 stars. Results sorted by stargazer count in descending order, limited to top 30 entries.

Validation Criteria: All repositories must have the ai-agent topic tag actively assigned. Repositories that removed or changed their topic tags between collection periods will appear as “dropped out” rather than losing stars.

Metrics Tracked: Repository name, star count, fork count, primary language, description, and last update timestamp.

Limitations: GitHub topic search does not reflect total star counts for repositories that lack the specific topic tag. A repository may retain high stars but drop from search results if the topic assignment changes.

This Week’s Data

Rank	Repository	Stars	Forks	Language	Description
1	NousResearch/hermes-agent	142,366	22,192	Python	The agent that grows with you
2	shareAI-lab/learn-claude-code	59,530	9,770	TypeScript	Bash is all you need — A nano claude code-like agent harness, built from 0 to 1
3	ruvnet/ruflo	48,438	5,367	TypeScript	The leading agent orchestration platform for Claude. Multi-agent swarms, autonomous workflows, native Claude Code integration
4	CherryHQ/cherry-studio	45,393	4,307	TypeScript	AI productivity studio with smart chat, autonomous agents, 300+ assistants
5	zhayujie/CowAgent	44,289	10,073	Python	CowAgent — LLM-based super AI assistant with multi-platform support
6	santifer/career-ops	43,911	9,214	JavaScript	AI-powered job search system built on Claude Code. 14 skill modes, Go dashboard
7	HKUDS/nanobot	42,142	7,400	Python	nanobot: The Ultra-Lightweight Personal AI Agent
8	CopilotKit/CopilotKit	31,216	4,037	TypeScript	The Frontend Stack for Agents & Generative UI. Makers of AG-UI Protocol
9	Gitlawb/openclaude	26,308	8,379	TypeScript	runs anywhere. uses anything
10	googleworkspace/cli	26,028	1,355	Rust	Google Workspace CLI — one CLI for Drive, Gmail, Calendar, Sheets, Docs. Includes AI agent skills
11	iOfficeAI/AionUi	24,334	2,196	TypeScript	Free local open-source 24/7 Cowork app for OpenClaw, Hermes Agent, Claude Code
12	nocobase/nocobase	22,334	2,589	TypeScript	AI + no-code platform for building business systems. WYSIWYG interface
13	activepieces/activepieces	22,135	3,635	TypeScript	AI Agents & MCPs & AI Workflow Automation. ~400 MCP servers for AI agents
14	jackwener/OpenCLI	19,583	1,924	JavaScript	Universal CLI Hub and AI-native runtime. AGENT.md integration for AI agents
15	Panniantong/Agent-Reach	19,134	1,700	Python	Give AI agent eyes to see the internet. Twitter, Reddit, YouTube, GitHub search
16	leon-ai/leon	17,218	1,443	TypeScript	Leon — open-source personal assistant
17	trycua/cua	15,869	994	HTML	Open-source infrastructure for Computer-Use Agents. Sandboxes, SDKs, benchmarks
18	hugohe3/ppt-master	14,246	1,439	Python	AI generates natively editable PPTX from any document
19	e2b-dev/E2B	12,130	902	Python	Open-source secure environment with real-world tools for enterprise agents
20	waooAI/waoowaoo	12,092	2,740	TypeScript	Industrial AI Agent platform for controllable film & video production
21	alibaba/OpenSandbox	10,531	843	Python	Secure, Fast, Extensible Sandbox runtime for AI agents
22	lsdefine/GenericAgent	10,490	1,194	Python	Self-evolving agent with skill tree, 6x less token consumption
23	frankbria/ralph-claude-code	9,074	688	Shell	Autonomous AI development loop for Claude Code with intelligent exit detection
24	EvoMap/evolver	7,352	748	JavaScript	GEP-powered self-evolving engine for AI agents. Genes, Capsules, Events
25	HKUDS/Vibe-Trading	6,761	1,339	Python	Vibe-Trading: Your Personal Trading Agent
26	crestalnetwork/intentkit	6,501	705	Python	Self-hosted cloud agent cluster with collaborative AI agent team
27	UfoMiao/zcf	5,977	419	TypeScript	Zero-Config Code Flow for Claude Code & Codex
28	Narcooo/inkos	5,961	1,121	TypeScript	Autonomous novel writing AI Agent with human review gates
29	holaboss-ai/holaOS	5,478	359	TypeScript	Open Agent Computer for ANY digital work
30	GetBindu/Bindu	5,424	363	Python	Turn any AI agent into a living microservice — interoperable, observable

Week-over-Week Summary

Metric	This Week (May 11)	Last Week (May 4)	Change
Total entries tracked	109	~100	+9 confirmed
Top 30 total stars	877,013	~743,128	+133,885 (+18.0%)
Average stars (top 30)	29,234	~24,771	+4,463 (+18.0%)
#1 Repository	Hermes Agent (142K)	AutoGPT (184K)	Complete changeover
Python repos (top 30)	11 (36.7%)	3 of prev top 5	Share decreased
TypeScript repos (top 30)	13 (43.3%)	1 of prev top 5	Now dominant
Repos dropped from top 5	—	5	100% turnover

Trends & Observations

Complete Leaderboard Transformation: The most striking finding this week is the total disappearance of all five previous top repositories from the ai-agent topic search results. AutoGPT (184K stars), Langflow (148K), Dify (140K), system-prompts-and-models (137K), and Langchain (136K) all vanished from the search — not due to star loss, but likely due to topic tag changes. This represents a 100% turnover at the top of the leaderboard.

Hermes Agent Emerges as Category Leader: NousResearch’s Hermes Agent now occupies the #1 position with 142,366 stars. The repository’s tagline “The agent that grows with you” suggests a focus on adaptive, learning-oriented agent architectures rather than static workflow automation.

Claude Code Ecosystem Dominance: At least five repositories in the current top 30 explicitly build on or integrate with Claude Code: learn-claude-code (#2), ruflo (#3), cherry-studio (#4), career-ops (#6), openclaude (#9), ralph-claude-code (#23), and zcf (#27). This clustering indicates a rapid ecosystem formation around Anthropic’s Claude Code tooling.

Language Distribution Shift: TypeScript now accounts for 43.3% of top 30 repositories (13 repos), surpassing Python at 36.7% (11 repos). This contrasts with the previous week where Python dominated the top positions. The shift reflects the developer tooling focus of Claude Code-compatible frameworks, which primarily target TypeScript/JavaScript environments.

MCP Integration Trend: activepieces (#13) explicitly highlights “~400 MCP servers for AI agents,” indicating that Model Context Protocol integration is becoming a standard feature for agent platforms. This aligns with the broader industry movement toward standardized agent-tool interfaces.

Self-Evolving Agent Architecture: Two repositories — GenericAgent (#22) with “skill tree” design and EvoMap/evolver (#24) with “GEP-powered self-evolving engine” — signal growing interest in agents that can autonomously expand their capabilities without manual configuration.

🔺 Scout Intel: What Others Missed

Confidence: high | Novelty Score: 85/100

The complete disappearance of AutoGPT, Langflow, Dify, system-prompts, and Langchain from the ai-agent topic search is not a coincidence — it signals a deliberate community and ecosystem shift. These repositories have either removed the ai-agent topic tag or GitHub’s topic indexing has fundamentally changed. Either explanation carries significant implications: if repository maintainers are reclassifying their projects away from “ai-agent,” it suggests the term has become too generic or stigmatized. If GitHub’s algorithm changed, it affects discoverability for an entire category. The simultaneous emergence of Claude Code-specific repositories at the top suggests a new ecosystem taxonomy centered around specific model providers rather than the broader “AI agent” abstraction.

Key Implication: Developers searching for “ai-agent” on GitHub will no longer discover the foundational frameworks (AutoGPT, Langchain) that defined the category in 2024-2025. This creates a discoverability gap for new entrants and positions Claude Code as the de facto standard for agent development in 2026 — a shift that benefits Anthropic’s ecosystem lock-in while fragmenting the broader open-source agent community.

Previous Snapshots

This is the first weekly snapshot in this tracker series. Historical snapshots will be listed here as they become available.

Sources

• GitHub Search API — ai-agent topic repositories — GitHub, 2026-05-11