The same REA ranking gives 234,477 and 347,463 stars from snapshots six days apart; the growth is not independently verified here.
Back to the trend of the day: A free tool lets AI take apart any app and copy its features
Poly Rating · from the trend of the day
GitHub Trending AI Tools 2026: Stars vs Installs
The same REA ranking gives 234,477 and 347,463 stars from snapshots six days apart; the growth is not independently verified here.
- No. 1: REA (morluto/rea)
- 10 places
01Final list
The leading AIs each built their own list of ten independently. Here is one result that is easy to check: every place has a reason, a source, and the list of AIs that included it.
Order: from most to least AI agreement.
A bright badge means this AI put the entry on its list. A dim badge means it did not.
- 1
REA (morluto/rea)
One answer cites 347,463 stars and 415,576 weekly npm downloads; another gives 300,000+ stars and 307,234 downloads, so both figures are disputed.
Included it: Claude, ChatGPT, Grok, Qwen. Did not include: Gemini, Meta.In their lists: Claude: 4 · ChatGPT: see list · Grok: 1 · Qwen: 9
What each AI noted 4
- Claudeplace 4
347,463 stars, the most in this list, yet 415,576 weekly npm downloads show real use; stars rose about 113,000 in the six days after a 14 Oct snapshot.
Source ↗ - ChatGPT
Named differently in its own list: open the full list.
see list - Grokplace 1
347k stars and +141k in one week, yet weekly npm downloads were ~232k—high attention, still early real installs relative to the star surge.
Source ↗ - Qwenplace 9
Gained 215,000 stars in one week, and actually converted that hype into 307,234 weekly npm downloads.
Source ↗
4 AIs put it here - 2
Cline (cline/cline)
One cited 20 October snapshot gives 70,706 stars and 42,575 weekly npm downloads; the npm figure misses extension installs.
Included it: Gemini, Claude, ChatGPT. Did not include: Meta, Grok, Qwen.In their lists: Gemini: 10 · Claude: 5 · ChatGPT: 4
What each AI noted 3
- Geminiplace 10
Has 70,000 stars but over 5 million marketplace installs, completely inverting the hype ratio into daily developer reliance.
Source ↗ - Claudeplace 5
70,706 stars but only 42,575 weekly npm downloads, because most users install the VS Code extension instead.
Source ↗ - ChatGPTplace 4
A tracker’s approximately 70,000 weekly downloads is the kind of install signal stars alone cannot supply.
3 AIs put it here - 3
OpenCode (anomalyco/opencode)
One answer pairs 218,854 stars with 2,123,309 weekly npm downloads; the figures come from its cited 20 October snapshot.
Included it: Claude, ChatGPT, Grok. Did not include: Gemini, Meta, Qwen.In their lists: Claude: 8 · ChatGPT: 1 · Grok: 3
What each AI noted 3
- Claudeplace 8
218,854 stars but 2.1 million weekly downloads, about one-sixth of Codex CLI's with 1.7 times the stars.
Source ↗ - ChatGPTplace 1
One independent live tracker lists about 213,000 weekly npm downloads alongside 129,000 GitHub stars: unusually strong evidence beyond bookmarks.
- Grokplace 3
~219k stars and strong npm demand (millions monthly); one of the better-aligned high-star tools.
Source ↗
3 AIs put it here - 4
Qwen Code (QwenLM/qwen-code)
A tracker cites about 89,000 stars and 28,000 weekly downloads; another cites about 27,000 stars and 45,040 downloads, so the comparison is unresolved.
Included it: Claude, ChatGPT, Grok. Did not include: Gemini, Meta, Qwen.In their lists: Claude: 7 · ChatGPT: 5 · Grok: 4
What each AI noted 3
- Claudeplace 7
A Gemini CLI fork with about 27k stars and 45,040 weekly downloads, a steadier ratio than the 2026 launches above.
Source ↗ - ChatGPTplace 5
Nearly 90,000 stars sit beside about 28,000 weekly downloads on one tracker, a visible gap between attention and installs.
- Grokplace 4
Lower stars than OpenCode in some snapshots yet higher real usage (39% of pros at work, tens of millions of installs).
Source ↗
3 AIs put it here - 5
OpenHands (All-Hands-AI/OpenHands)
A cited tracker reports about 96,000 stars, while another answer gives 91,400; the snapshots differ, and neither provides a comparable install count.
Included it: Gemini, Claude, ChatGPT. Did not include: Meta, Grok, Qwen.In their lists: Gemini: 8 · Claude: see list · ChatGPT: 2
What each AI noted 3
- Geminiplace 8
Maintains 91,000 stars with 3.4M+ container pulls, proving that star hype can convert into enterprise infrastructure.
Source ↗ - Claude
Named differently in its own list: open the full list.
see list - ChatGPTplace 2
It combines a roughly 96,000-star tracker count with a separate survey signal that measures adoption rather than GitHub enthusiasm.
3 AIs put it here - 6
Gemini CLI (google-gemini/gemini-cli)
One tracker cites about 51,000 stars and 107,000 weekly downloads; another says the tool was retired but still had 227,214 downloads in a later week.
Included it: Gemini, Claude, ChatGPT. Did not include: Meta, Grok, Qwen.In their lists: Gemini: see list · Claude: see list · ChatGPT: 7
3 AIs put it here - 7
DeepSeek Harness (deepseek-ai/deepseek-harness)
One answer reports 255,733 stars and 136 weekly downloads for a same-named npm package; it warns that the package may not be the main install channel.
Included it: Claude, Grok, Qwen. Did not include: Gemini, Meta, ChatGPT.In their lists: Claude: 2 · Grok: see list · Qwen: 3
What each AI noted 3
- Claudeplace 2
255,733 stars in about 68 days against 136 weekly downloads of the npm package of the same name.
Source ↗ - Grok
Named differently in its own list: open the full list.
see list - Qwenplace 3
Crossed 100,000 stars in just 53 hours, but developers mostly install its standalone Desktop app, not a registry package.
Source ↗
3 AIs put it here - 8
Kilo Code (Kilo-Org/kilocode)
One tracker pairs about 88,000 stars with 28,000 weekly downloads; another reports about 27,000 stars and 16,095 CLI downloads.
Included it: Claude, ChatGPT. Did not include: Gemini, Meta, Grok, Qwen.In their lists: Claude: 6 · ChatGPT: 6
What each AI noted 2
- Claudeplace 6
About 27k stars (29 Aug snapshot) against 16,095 weekly CLI downloads, as Roo Code's former users migrate.
Source ↗ - ChatGPTplace 6
About 28,000 weekly downloads against roughly 88,000 stars puts it in the middle of the attention-to-install spread.
2 AIs put it here - 9
OpenAI Codex CLI (openai/codex)
One answer reports 130,164 stars and 13,153,561 weekly npm downloads; the download count is a single dated snapshot, not verified against a live page here.
Included it: Claude. Did not include: Gemini, Meta, ChatGPT, Grok, Qwen.In their lists: Claude: 10
What each AI noted 1
- Claudeplace 10
13.2 million weekly downloads on 130,164 stars: attention that turned into installs, about 100,000 downloads per 1,000 stars.
Source ↗
1 AIs put it here - 10
claw-code (ultraworkers/claw-code)
One answer reports 194,950 stars and 107,890 forks, but found no npm package with the same name; the apparent gap may reflect another distribution path.
Included it: Claude. Did not include: Gemini, Meta, ChatGPT, Grok, Qwen.In their lists: Claude: 1
What each AI noted 1
- Claudeplace 1
194,950 stars and 107,890 forks (about 55 forks per 100 stars) for a repo that calls itself an 'agent-managed museum exhibit'.
Source ↗
1 AIs put it here
02Worth knowing
What the AIs noticed while building the list.
Weekly downloads are not unique users and may include CI, mirrors, and repeat installs; npm-only counts can miss editor extensions.
The fake-star study estimates suspicious activity across GitHub, but does not provide a verified fake-star share for each listed tool.
03How we counted
In plain words, no formulas.
1The question
Top 10 AI Developer Tools of 2026 Where GitHub Stars and Real Installs Differ Most
2Why this topic
The reader can see which hyped tools developers actually install. The trending page and the download counts each tell only half of that story.
3Where it applies
Global, open-source AI coding-agent tools released or trending in 2026
4What we looked at
- Star count and star growth in the first weeks after launch
- Weekly npm or PyPI downloads of the same tool
- Repository age compared with the speed of its star growth
- Share of stars from suspicious or empty accounts, where a study or tool has measured it
- GitHub repository pages and API — Current stars and repository age; snapshots change over time, and stars measure attention rather than installs.
- npm download API — Weekly package downloads; counts can include repeat installs, CI, and mirrors and are not unique users.
- StarScout study — Estimates suspicious star activity across GitHub; it does not establish a verified suspicious-star share for each tool here.
5Who ranked
The leading AIs, each building its own list of ten independently and naming a source for every place: Gemini, Claude, Meta, ChatGPT, Grok, Qwen.
6How the places were set
The same entry written differently by different AIs was counted as one. The more AIs put an entry on their list, and the higher they put it, the higher its place. "AI agreement" shows how many of the AIs had it on their own list. Sources are only those the AIs found themselves; nothing was added by hand.
04Each AI's own list
🇺🇸 Gemini10 places
The same question, answered by Gemini alone. The final list above combines all the AIs.
- 1AutoGPT
Amassed over 165,000 GitHub stars while its daily active workflow runs plunged by over 80% once practical coding tasks failed.
Source ↗ - 2morluto/rea
Shot to No. 1 on GitHub Trending with tens of thousands of stars in October 2026 while npm package downloads hovered under 3,500.
Source ↗ - 3gpt-engineer
Gained 52,000 stars on the promise of one-prompt full applications, but active command-line installs trail by orders of magnitude.
Source ↗ - 4ChatDev
Virtual software company simulator gained 26,000 stars despite near-zero production deployments in commercial dev environments.
Source ↗ - 5MetaGPT
Boasts 45,000 stars for multi-agent software generation while actual enterprise workflow imports remain under 4% of stargazers.
Source ↗ - 6Devv AI
Rocketed on GitHub trending lists through social promotions while real package utilization remained confined to web search tabs.
Source ↗ - 7GPT Researcher
Gained 19,000 stars for autonomous web scraping agents while active API key users represent a small fraction of bookmarks.
Source ↗ - 8OpenHands (All-Hands-AI)
Maintains 91,000 stars with 3.4M+ container pulls, proving that star hype can convert into enterprise infrastructure.
Source ↗ - 9Roo Code
Community fork surpassed 20,000 stars with real extension installs exceeding 350,000 on the VS Code Marketplace.
Source ↗ - 10Cline
Has 70,000 stars but over 5 million marketplace installs, completely inverting the hype ratio into daily developer reliance.
Source ↗
🇺🇸 Claude10 places · 4 notes
The same question, answered by Claude alone. The final list above combines all the AIs.
- 1claw-code (ultraworkers/claw-code, 2026)
194,950 stars and 107,890 forks (about 55 forks per 100 stars) for a repo that calls itself an 'agent-managed museum exhibit'.
Source ↗ - 2DeepSeek Harness (deepseek-ai/deepseek-harness, 2026)
255,733 stars in about 68 days against 136 weekly downloads of the npm package of the same name.
Source ↗ - 3ECC, 'agent harness performance optimization system' (affaan-m/ECC, 2026)
284,864 stars after nine months, against 9,377 weekly downloads for the npm package I tried (ecc-universal).
Source ↗ - 4REA, Reverse Engineer Anything (morluto/rea, 2026)
347,463 stars, the most in this list, yet 415,576 weekly npm downloads show real use; stars rose about 113,000 in the six days after a 14 Oct snapshot.
Source ↗ - 5Cline (cline/cline)
70,706 stars but only 42,575 weekly npm downloads, because most users install the VS Code extension instead.
Source ↗ - 6Kilo Code (Kilo-Org/kilocode)
About 27k stars (29 Aug snapshot) against 16,095 weekly CLI downloads, as Roo Code's former users migrate.
Source ↗ - 7Qwen Code (QwenLM/qwen-code)
A Gemini CLI fork with about 27k stars and 45,040 weekly downloads, a steadier ratio than the 2026 launches above.
Source ↗ - 8OpenCode (anomalyco/opencode)
218,854 stars but 2.1 million weekly downloads, about one-sixth of Codex CLI's with 1.7 times the stars.
Source ↗ - 9Pi coding agent (earendil-works/pi)
116,700 stars and 2.7 million weekly downloads, among the best conversion rates in the open-source group.
Source ↗ - 10OpenAI Codex CLI (openai/codex)
13.2 million weekly downloads on 130,164 stars: attention that turned into installs, about 100,000 downloads per 1,000 stars.
Source ↗
What Claude noticed
Codex CLI has 13,153,561 weekly npm downloads on 130,164 stars. OpenCode has 2,123,309 downloads on 218,854 stars. The tool with fewer stars is installed about six times more.
claw-code has 107,890 forks on 194,950 stars (about 55%), against about 16% for Codex (20,505 forks on 130,164 stars). Its README calls it a museum exhibit rather than a product.
The ICSE 2026 StarScout study found about 6 million suspected fake stars. AI and LLM repos were the largest non-malicious category of fake-star recipients, but the effect on rankings lasts less than two months.
Gemini CLI was retired for most users on 18 June 2026 (per Pinggy), yet its npm package still logged 227,214 downloads in the week of 12-18 Oct.
🇺🇸 ChatGPT10 places · 2 notes
The same question, answered by ChatGPT alone. The final list above combines all the AIs.
- 1OpenCode
One independent live tracker lists about 213,000 weekly npm downloads alongside 129,000 GitHub stars: unusually strong evidence beyond bookmarks.
- 2OpenHands
It combines a roughly 96,000-star tracker count with a separate survey signal that measures adoption rather than GitHub enthusiasm.
- 3OpenAI Codex CLI
Its 2026 adoption grew about fivefold in the JetBrains report, while the npm package count still needs careful interpretation.
Source ↗ - 4Cline
A tracker’s approximately 70,000 weekly downloads is the kind of install signal stars alone cannot supply.
- 5Qwen Code
Nearly 90,000 stars sit beside about 28,000 weekly downloads on one tracker, a visible gap between attention and installs.
- 6Kilo Code
About 28,000 weekly downloads against roughly 88,000 stars puts it in the middle of the attention-to-install spread.
- 7Gemini CLI
A tracker lists roughly 107,000 weekly downloads, while JetBrains’ adoption measure gives a different view of reach.
- 8Goose
Roughly 55,000 weekly downloads accompany only about 51,000 stars in one snapshot, a relatively install-heavy profile.
- 9Claude Code
The tracker places it near 150,000 weekly downloads, but this does not capture its full usage because the npm package is only one install path.
- 10Aider
About 50,000 stars and 100 monthly commits on one tracker show why accumulated stars can outlive current momentum.
What ChatGPT noticed
JetBrains reports Codex adoption rising from 3% in January 2026 to 16% in May–July, a survey measure that tells a different story from GitHub stars.
A 2026 global study estimates six million suspected fake GitHub stars overall and finds campaigns increasingly target AI/LLM repositories; it does not establish that any specific tool in this ranking bought stars.
🇺🇸 Grok10 places · 4 notes
The same question, answered by Grok alone. The final list above combines all the AIs.
- 1REA (morluto/rea) 2026
347k stars and +141k in one week, yet weekly npm downloads were ~232k—high attention, still early real installs relative to the star surge.
Source ↗ - 2mattpocock/skills (agent skills)
Hundreds of thousands of stars and +1.5k/day recently; practical installs lag far behind pure attention.
Source ↗ - 3OpenCode (anomalyco/opencode)
~219k stars and strong npm demand (millions monthly); one of the better-aligned high-star tools.
Source ↗ - 4Claude Code (anthropics/claude-code)
Lower stars than OpenCode in some snapshots yet higher real usage (39% of pros at work, tens of millions of installs).
Source ↗ - 5Ponytail / similar prompt-skill tools
Fastest-growing in some reports (+34k/30 days) with limited evidence of sustained package installs.
Source ↗ - 6OpenAI Codex CLI
~130k stars and extremely high npm downloads (93M cited monthly in one index)—usage exceeds pure star narrative.
- 7n8n
~208k stars with solid but lower relative npm volume compared with pure coding-agent spikes.
- 8LangGraph / LangChain ecosystem
High stars and strong PyPI downloads; more mature adoption than pure 2026 hype tools.
- 9VoiceStudio and similar media agents
Rapid star gains (+2k/day) with less transparent install data; typical of attention-first tools.
Source ↗ - 10paperclip and agent-orchestration tools
Moderate star velocity with sparse public download figures; high bookmark-to-use ratio likely.
Source ↗
What Grok noticed
REA’s star explosion (+141k/week) far outpaced its early npm installs, illustrating classic hype-to-use lag.
OpenCode often leads stars while Claude Code leads actual workplace usage surveys.
A 2026 ICSE paper estimated ~6 million fake stars, with AI/LLM repos among the most affected.
Stars measure one click; npm/PyPI downloads require an actual install and better track sustained use.
🇨🇳 Qwen10 places · 3 notes
The same question, answered by Qwen alone. The final list above combines all the AIs.
- 1OpenClaw
Beat React's 10-year GitHub star record in 60 days, yet registers zero npm or PyPI downloads because it installs via Git or standalone binaries.
Source ↗ - 2Hermes Agent
Sitting at over 250,000 stars, its actual footprint is just 53,000 weekly PyPI downloads, showing a massive gap between bookmarks and active use.
Source ↗ - 3DeepSeek Harness
Crossed 100,000 stars in just 53 hours, but developers mostly install its standalone Desktop app, not a registry package.
Source ↗ - 4Awesome / Curated AI Tool Lists
Repos like awesome-ai-tools collect hundreds of stars daily as bookmarks, while the tools they link to see the actual installs.
Source ↗ - 5Fake AI Repos (Star Farms)
Star farms push thousands of fake stars to lookalike repos to steal credentials, but CI logs show zero real downloads.
Source ↗ - 6usestrix/strix
Trended at #1 in July 2026 with ~42k stars, but mostly used as a bookmark for AI security research rather than daily installs.
Source ↗ - 7Grok Build (xai-org/grok-build)
Grabbed 8,700 stars in 48 hours, but installs happen via native macOS/Windows binaries, bypassing package registries entirely.
Source ↗ - 8PocketPaw
Marketed as a one-command OpenClaw alternative for laptops, translating high GitHub interest into actual PyPI downloads.
Source ↗ - 9REA (Reverse Engineer Anything)
Gained 215,000 stars in one week, and actually converted that hype into 307,234 weekly npm downloads.
Source ↗ - 10LangChain
The baseline: 132,000 stars and 53 million weekly PyPI downloads show perfect alignment between attention and massive production use.
Source ↗
What Qwen noticed
CMU researchers found that nearly 17% of all GitHub repos with 50+ stars were targeted by fake star campaigns by mid-2024.
OpenClaw broke GitHub's star records, yet its actual adoption happens outside npm/PyPI via direct Git clones and Docker containers.
Star farms charge as little as $0.03 to $0.90 per star, making it cheap to trend on GitHub without writing a single line of useful code.