This Week in Open-Source AI Agents — Top Trending GitHub Repos (Week of June 29, 2026)
Hand-edited weekly roundup of top trending open-source AI agent, LLM, and MCP projects on GitHub — data auto-collected by Dibi8 Tribe Intel, analysis by Dibi8 editorial team.
- Updated 2026-06-29
TL;DR: This week’s trending repos span AI video editing, cybersecurity skills for agents, open-source design tools, AI website cloning, and privacy-first messaging. Palmier Pro leads with 6,126 stars/week as the first AI-native macOS video editor.
Why We Watch This Week #
Open-source AI moves fast. This week’s trending repos show a clear pattern: AI agents are expanding beyond coding into creative tools, security, and privacy infrastructure. The biggest gainers aren’t LLM frameworks — they’re domain-specific applications that give agents real capabilities.
What makes W26 particularly interesting is the diversity of domains represented. We have a macOS-native video editor (Palmier Pro), a massive cybersecurity skill library (Anthropic Skills), a mature open-source design platform (Penpot), an AI-powered website cloning tool, and a privacy-first messaging protocol (SimpleX). Together, these repos paint a picture of an ecosystem maturing from experimental prototypes to production-ready tools.
Here are the 5 most noteworthy repos from GitHub Trending (weekly) that haven’t been covered in previous editions. from GitHub Trending (weekly) that haven’t been covered in previous editions.
1. Palmier Pro — AI Video Editor for macOS #
Stars this week: 6,126 | Total stars: 9,295 | Language: Swift | License: GPL-3.0 Topics: ai-video, claude, macos, mcp, seedance2, swift, video-editor
What It Is #
Palmier Pro is the first video editor built specifically for AI agents. Unlike traditional editors that bolt on AI features as plugins, Palmier Pro’s entire architecture assumes AI is the primary editing interface.
The app integrates with Claude and Seedance 2 for AI-powered video generation, editing, and post-production. Its MCP (Model Context Protocol) support means AI agents can directly manipulate video timelines, apply effects, and generate content — no manual UI interaction required.
Why It Matters #
Video creation is one of the fastest-growing AI application areas. Palmier Pro represents a shift from “AI-assisted editing” to “AI-native editing” — where the editor understands intent rather than requiring manual timeline manipulation.
For developers and content creators, this means the ability to describe what you want in natural language and have an AI agent construct the video, apply transitions, and export — all within a professional-grade macOS application.
Hands-On Notes #
The requirement for macOS 26 (Tahoe) suggests this is a cutting-edge release targeting Apple’s latest platform features. The Swift implementation means tight integration with Metal for GPU-accelerated video processing.
2. Anthropic Cybersecurity Skills — 817 Structured Skills for AI Agents #
Stars this week: 5,121 | Total stars: 22,612 | Language: Python | License: Apache-2.0 Topics: ai-agents, claude-code, cloud-security, cybersecurity, devsecops, ethical-hacking, incident-response, infosec, llm, malware-analysis, mcp, mitre-attack, nist-csf, osint, penetration-testing, red-team, security, security-automation, threat-hunting, threat-intelligence
What It Is #
The largest open-source cybersecurity skills library for AI agents, featuring 817 structured skills mapped to major security frameworks including MITRE ATT&CK, NIST CSF, and various red team methodologies.
Each skill is a structured prompt or tool definition that enables AI agents to perform specific cybersecurity tasks — from vulnerability scanning and malware analysis to incident response and threat hunting.
Why It Matters #
This represents a paradigm shift in how security professionals work with AI. Instead of manually running security tools, agents equipped with these skills can:
- Autonomously scan infrastructure for vulnerabilities
- Perform structured penetration testing with MITRE-mapped techniques
- Conduct incident response with predefined playbooks
- Generate threat intelligence reports from raw data
- Run security assessments against compliance frameworks
For the dibi8 audience interested in AI agents, this is a goldmine — it demonstrates how specialized domain knowledge can be encoded as agent skills that anyone can deploy.
Hands-On Notes #
With 22,612 total stars and 2,578 forks, this repo has achieved significant community traction. The Apache-2.0 license makes it suitable for commercial integration. The breadth of topics (19 distinct security domains) means this isn’t a niche tool — it’s a comprehensive security capability library.
3. Penpot — Open-Source Design Tool for Teams #
Stars this week: 3,343 | Total stars: 54,379 | Language: Clojure | License: MPL-2.0 Topics: clojure, clojurescript, design, prototyping, ui, ux-design, ux-experience
What It Is #
Penpot continues its meteoric rise as the leading open-source alternative to Figma. This week’s surge of 3,343 new stars reflects growing demand for self-hosted design tools, particularly among teams concerned about data sovereignty and vendor lock-in.
Penpot supports real-time design-to-code collaboration, allowing designers and developers to work on the same canvas. Its web-based architecture means it runs anywhere — browser, self-hosted server, or cloud deployment.
Why It Matters #
The design tool market is dominated by Figma (Adobe), and Penpot is the only serious open-source competitor. With 54,379 total stars, it has proven its viability and community support.
For organizations that need design tools without cloud dependency, Penpot offers:
- Self-hosted deployment — full control over design data
- Design-to-code workflow — developers get clean CSS/SVG output
- Real-time collaboration — multiple designers working simultaneously
- Open file format — no vendor lock-in for design assets
- AI-ready architecture — extensible plugin system for AI integrations
Hands-On Notes #
Penpot’s Clojure/ClojureScript stack is unusual for a design tool, but it delivers excellent performance and a robust type system. The 382MB repository size reflects a mature, feature-complete product. With 667 open issues, the community is actively contributing improvements.
→ Try Penpot Online | Self-Host Documentation
Quick Start: Self-Hosting Penpot #
Penpot provides a Docker Compose setup for self-hosting:
version: "3.6"
services:
penpot-backend:
image: penpotapp/backend:latest
ports:
- 9001:9001
environment:
- PENPET_PUBLIC_URI=http://localhost:9000
- PENPET_SECRET_KEY=penpot-secrets-change-me
penpot-frontend:
image: penpotapp/frontend:latest
ports:
- 9000:80
environment:
- PENPET_PUBLIC_URI=http://localhost:9000
- PENPET_BACKEND_URI=http://localhost:9001
After deployment, access Penpot at http://localhost:9000 and begin creating designs with your team.
4. AI Website Cloner Template — One-Command Website Duplication #
Stars this week: 4,565 | Total stars: TBD | Language: TypeScript | License: TBD Topics: ai-coding, website-cloning, template-generation
What It Is #
A tool that clones any website into a structured project template using AI coding agents. One command, and you get a clean, editable codebase that mirrors the target website’s structure, styling, and content.
Unlike scraping tools that produce messy HTML, this tool leverages AI to produce clean, maintainable code — complete with component separation, proper CSS architecture, and semantic HTML structure.
Why It Matters #
Website cloning has a reputation for being used for phishing and copyright infringement. However, the legitimate use cases are substantial:
- Design inspiration analysis — study how competitors structure their sites
- Legacy migration — modernize old websites with AI-generated clean code
- Educational purposes — learn web development by examining real-world examples
- Portfolio reconstruction — rebuild your own archived websites
For AI coding enthusiasts, this represents a practical application of AI agents that goes beyond code generation into full-project understanding.
Hands-On Notes #
The TypeScript implementation suggests Node.js-based tooling, likely integrating with popular AI coding frameworks. At 4,565 stars/week, it’s clearly hitting a nerve in the developer community.
5. SimpleX Chat — Privacy-First Messaging Without User IDs #
Stars this week: 1,973 | Total stars: TBD | Language: Haskell | License: TBD Topics: privacy, messaging, decentralization, cryptography
What It Is #
SimpleX is the first messaging network that operates completely without user identifiers. No phone numbers, no email addresses, no usernames — just anonymous connections that preserve privacy by design.
Traditional messaging apps (Signal, WhatsApp, Telegram) all require some form of persistent identity. SimpleX eliminates this entirely by using ephemeral connection addresses that change with each conversation.
Why It Matters #
Privacy is increasingly important in an era of mass surveillance and data breaches. SimpleX’s approach is fundamentally different from “encrypted but identifiable” messengers:
- No persistent identity — each conversation uses a unique, disposable connection
- No metadata collection — the network doesn’t know who talks to whom
- Decentralized architecture — no central server controls your conversations
- Open-source implementation — full transparency and auditability
For developers interested in privacy-preserving technologies, SimpleX represents cutting-edge research in anonymous communication protocols.
Hands-On Notes #
Built in Haskell, SimpleX leverages functional programming’s strengths in formal verification and mathematical correctness — crucial for security-critical applications. The 1,973 stars/week growth indicates strong interest in privacy-first alternatives.
This Week’s Trends #
Looking at W26’s trending repos, three patterns emerge:
Domain-specific AI agents — From video editing (Palmier Pro) to cybersecurity (Anthropic Skills), AI agents are becoming specialized tools rather than general-purpose assistants.
Open-source infrastructure maturity — Penpot’s 54K+ stars and consistent weekly growth show that open-source alternatives to commercial tools are reaching parity and gaining adoption.
Privacy-first design — SimpleX’s no-ID approach represents a philosophical shift: privacy shouldn’t be a feature you opt into; it should be the default architecture.
Looking Ahead #
Next week, we’ll be watching for:
- AI video tools — Palmier Pro’s growth suggests more competitors may emerge in the AI-native video editing space
- Security skills libraries — The success of 817 cybersecurity skills indicates demand for domain-specific agent capabilities
- Privacy infrastructure — SimpleX’s approach challenges conventional wisdom about user identity in messaging
The open-source AI ecosystem is moving from “can we build it?” to “should we build it?” — and that’s a sign of maturity.
How We Collect This Data #
Dibi8 Tribe Intel uses an automated script to scrape GitHub Trending (weekly), cross-reference with our existing article database to avoid duplicates, and verify image assets via README analysis. Human editors then select the top 5 based on relevance, star velocity, and potential value to our readers.
All data is sourced from GitHub’s public API and trending pages. Star counts are verified via API to avoid the discrepancies between trending page estimates and actual counts.
Frequently Asked Questions #
Q: How does Dibi8 select repos for the weekly trending report?
A: We use an automated script that scrapes GitHub Trending (weekly), cross-references with our existing article database to avoid duplicates, and verifies image assets via README analysis. Human editors then select the top 5 based on relevance to our AI developer audience, star velocity, and potential educational value.
Q: Why do some repos have different star counts on GitHub vs. this report?
A: GitHub Trending shows approximate weekly star gains, while our report verifies exact counts via the GitHub API. The API provides the authoritative total, while the trending page estimate may lag or be inflated by viral spikes.
Q: Can I suggest a repo for next week’s trending report?
A: Yes! Submit repos via our GitHub discussions or reach out on our community channels. We review all suggestions and add them to our candidate pool.
Q: How often is this report published?
A: Every Monday. Data is collected from GitHub Trending at the time of publication, reflecting repos that gained the most stars in the preceding 7-day period.
Q: Are there affiliate links in this article?
A: No. Dibi8 maintains strict editorial independence. All links point to official GitHub repositories or project websites. We do not accept payment for inclusion in trending reports.
More from Dibi8 #
- Open-Source AI Tools Directory — Browse our complete collection of AI tools, frameworks, and resources.
- AI Agent Tool Chain — Curated collection of AI agent development tools.
- Weekly Trending Archives — Previous weeks’ trending reports.
This Week in Open-Source AI Agents is published weekly by Dibi8 Tribe Intel. Data collected: June 29, 2026. Next edition: July 6, 2026.
💬 Discussion