Updated weekly
Week 39 · 2026
Top 10 AI agent and automation repos on GitHub, week 39 2026
Stars gained 21 Sept–27 Sept · New podium every Monday
hindsight
Hindsight gives an AI agent a memory that gets smarter over time, instead of forgetting everything between chats. It's a tool developers plug into their own AI agent or chatbot, not something you install yourself. If your business already has a custom AI assistant, ask your developer if a memory tool like this would help.
Why it's trending: Independent tests by researchers at Virginia Tech and The Washington Post found it is the most accurate agent-memory tool on a widely used benchmark called LongMemEval.
- Stars this week
- +7,282
- Total stars
- 37,121
- Language
- Python
- License
- MIT
- Can a non-developer use it?
- Developers only
paperclip
Paperclip lets you set up a small team of AI agents and give them business goals to work on. One dashboard tracks their work and cost. It's for people already using AI coding agents like Claude Code, not a typical shop or clinic. Setting it up needs a command line and API keys from an AI provider.
Why it's trending: It gained 5,376 stars this week.
- Stars this week
- +5,376
- Total stars
- 89,637
- Language
- TypeScript
- License
- MIT
- Can a non-developer use it?
- With help
orca
Orca is a desktop app that runs several AI coding agents side by side, each in its own copy of the code, with a phone app to check on them remotely. It's built for developers who manage AI coding work, not for a non-technical business owner.
Why it's trending: It gained 4,696 stars this week.
- Stars this week
- +4,696
- Total stars
- 79,501
- Language
- TypeScript
- License
- MIT
- Can a non-developer use it?
- Developers only
Ranks 4 to 10
ECCECC helps manage several AI coding agents running at once. It adds shared memory, safety limits, and monitoring across tools like Claude Code and Codex. It's built for software teams, so it only matters if your developers already use AI coding agents.Developers only+3,560 ★
ECC helps manage several AI coding agents running at once. It adds shared memory, safety limits, and monitoring across tools like Claude Code and Codex. It's built for software teams, so it only matters if your developers already use AI coding agents.
Why it's trending: It gained 3,560 stars this week.
Total stars: 268,362 · Language: JavaScript
affaan-m/ECC · View on GitHub ↗security-audit-skillThis skill turns your AI coding agent into a security auditor for your code. It hunts for real security flaws in stages, then has a second AI agent double-check each finding. It's built for developers — if a freelancer built your website, ask if they can run a check like this.Developers only+3,214 ★
This skill turns your AI coding agent into a security auditor for your code. It hunts for real security flaws in stages, then has a second AI agent double-check each finding. It's built for developers — if a freelancer built your website, ask if they can run a check like this.
Why it's trending: It gained 3,214 stars this week.
Total stars: 22,308 · Language: JavaScript
cloudflare/security-audit-skill · View on GitHub ↗open-code-reviewOpen Code Review is a command-line tool that uses an AI model to check code changes before they go live. It started as Alibaba's own internal review tool, and the company later made it free and open. It's for developers, but shows how AI code review is becoming normal for the people who build your software.Developers only+2,614 ★
Open Code Review is a command-line tool that uses an AI model to check code changes before they go live. It started as Alibaba's own internal review tool, and the company later made it free and open. It's for developers, but shows how AI code review is becoming normal for the people who build your software.
Why it's trending: It gained 2,614 stars this week.
Total stars: 41,905 · Language: Go
alibaba/open-code-review · View on GitHub ↗impeccableImpeccable gives an AI coding agent design rules, so its websites and apps don't all end up looking the same. It plugs into tools like Claude Code and checks a design against dozens of rules automatically. It's a tool for developers and the freelancers who build your website, not something a business owner installs.Developers only+2,448 ★
Impeccable gives an AI coding agent design rules, so its websites and apps don't all end up looking the same. It plugs into tools like Claude Code and checks a design against dozens of rules automatically. It's a tool for developers and the freelancers who build your website, not something a business owner installs.
Why it's trending: It builds on Anthropic's own frontend-design skill for Claude, adding a much bigger set of design checks.
Total stars: 71,798 · Language: JavaScript
pbakaus/impeccable · View on GitHub ↗nasikoNasiko controls many AI agents that talk to each other over a shared standard called A2A. It deploys, routes calls to, and watches over these agents from one command. It's built for development teams running several AI agents, not for someone without technical experience.Developers only+2,342 ★
Nasiko controls many AI agents that talk to each other over a shared standard called A2A. It deploys, routes calls to, and watches over these agents from one command. It's built for development teams running several AI agents, not for someone without technical experience.
Why it's trending: It gained 2,342 stars this week.
Total stars: 8,996 · Language: Rust
Nasiko-Labs/nasiko · View on GitHub ↗WeKnoraWeKnora turns your documents, like PDFs and Word files, into a searchable knowledge base an AI agent can answer from. A business could use it to build an assistant that answers staff questions from your own files. It needs a technical person to set up with a tool called Docker, and it stays on your own servers.With help+2,007 ★
WeKnora turns your documents, like PDFs and Word files, into a searchable knowledge base an AI agent can answer from. A business could use it to build an assistant that answers staff questions from your own files. It needs a technical person to set up with a tool called Docker, and it stays on your own servers.
Why it's trending: It gained 2,007 stars this week.
Total stars: 30,583 · Language: Go
Tencent/WeKnora · View on GitHub ↗financial-servicesClaude for Financial Services is a set of ready-made AI agents for finance work. They draft pitch decks, build valuation models, and check accounting numbers. It's built for banks, investment firms, and finance teams, not a typical small shop. Every agent drafts work for a person to check; none can approve a deal or move money on its own.Yes+1,969 ★
Claude for Financial Services is a set of ready-made AI agents for finance work. They draft pitch decks, build valuation models, and check accounting numbers. It's built for banks, investment firms, and finance teams, not a typical small shop. Every agent drafts work for a person to check; none can approve a deal or move money on its own.
Why it's trending: It gained 1,969 stars this week.
Total stars: 37,852 · Language: Python
anthropics/financial-services · View on GitHub ↗
● Updated weekly · Next podium: Monday 5 Oct
Ranked by stars gained on GitHub this week, filtered for AI agent and automation projects that do real work. We leave out lists, courses and joke repos.
See this week's podium →