AI Open Source Projects · July 21, 2026
AI Open Source Projects · July 21, 2026
Weekly selection of notable AI open source projects.
1. Omnigent ★ 7.6k
Open-source AI agent framework and meta-harness: orchestrate Claude Code, Codex, Cursor, and other AI coding agents in coordinated workflows. Omnigent provides a unified interface for managing multiple agent types, routing tasks between them, and aggregating results. Built for teams building complex multi-agent systems.
GitHub: https://github.com/omnigent-ai/omnigent
2. T3MP3ST ★ 5.1k
Autonomous red teaming platform by elder-plinius. T3MP3ST is a multi-agent offensive-security meta-harness designed to stress-test AI systems through coordinated adversarial attacks. It simulates sophisticated threat actors to identify vulnerabilities in LLM-powered applications before real attackers do.
GitHub: https://github.com/elder-plinius/T3MP3ST
3. OpenScience ★ 2.7k
The open-source AI workbench for scientific research, by synthetic-sciences. OpenScience provides a local-first, model-agnostic environment for scientists to leverage AI across the research lifecycle: literature review, hypothesis generation, experimental design, data analysis, and paper writing. Supports multiple LLM backends.
GitHub: https://github.com/synthetic-sciences/openscience
4. SuperLog ★ 1.0k
Open-source observability tool that uses AI agents to self-heal your software. SuperLog ingests logs, metrics, and traces, then employs specialized AI agents to detect anomalies, diagnose root causes, and even propose or apply fixes automatically. Think of it as an AI-powered SRE on call.
GitHub: https://github.com/superloglabs/superlog
5. Sandboxd ★ 852
Open-source, self-hosted AI app builder. An agent builds real applications in isolated sandboxes on your own infrastructure. Describe your app idea in natural language, and Sandboxd spins up a secure sandbox, generates the code, and deploys it. Supports multiple frameworks and languages.
GitHub: https://github.com/tastyeffectco/sandboxd
6. Talos ★ 988
GPU worker client for the Talos network. Pairs with your Talos account to serve open-model inference. Talos is a decentralized GPU network that lets model owners deploy open-weight models and earn rewards, while users get affordable access to a wide range of open models without centralized cloud providers.
GitHub: https://github.com/jmerelnyc/Talos
7. VulnHunter ★ 610
Agentic AI security tool by Capital One that applies proactive, attacker-first analysis directly to source code. VulnHunter uses AI agents to simulate attacker techniques, identify vulnerabilities, and suggest fixes. Designed for integration into CI/CD pipelines for continuous security assessment.
GitHub: https://github.com/capitalone/VulnHunter
8. kill-ai-slop ★ 675
A field guide to the visual and copy tics of AI-generated products, plus an Agent Skill that scans your codebase for them. Helpful for teams that want to ensure their AI-assisted output maintains a human touch and avoids the telltale patterns of AI-generated content.
GitHub: https://github.com/yetone/kill-ai-slop
Curated from GitHub Trending and Hacker News.
