Best Open Source products of the month May 2026

Radar

The missing open-source Kubernetes UI

Radar consolidates your entire Kubernetes workflow into one exceptionally fast, open-source user interface. Gain unparalleled visibility with real-time topology mapping, resource utilization tracking, event monitoring, Helm chart management, and robust GitOps integration. It offers live traffic flow analysis, comprehensive security and best-practice checks, in-depth image filesystem inspection, and Multi-Cloud Protocol (MCP) support for seamless integration with AI agents. Deploy Radar effortlessly, either locally as a single binary or self-host it within your cluster with full RBAC and OIDC support, eliminating the need for external accounts, agents, or cloud dependencies. This powerful solution provides a complete operational overview for enhanced control and efficiency.

BugDrop

In-app feedback that creates GitHub Issues with screenshots

Bugdrop-2 is a free, open-source website feedback widget designed to simplify bug reporting. It allows users to effortlessly report issues directly from your website, complete with screenshots and precise annotations. These reports are then automatically converted into actionable issues within GitHub, streamlining your development workflow. Integration is remarkably simple, requiring just one script tag with zero configuration, and it's fully customizable to seamlessly align with your application's branding and design.

Agentmemory

Persistent memory for Claude Code, Codex & coding agents

Unlock unprecedented capabilities for your AI agents by providing Hermes, Claude Code, and Codex with truly infinite memory. Agentmemory is rapidly gaining recognition, highlighted by its trending status on GitHub with over 5,000 stars, demonstrating its impact within the developer community. This innovative solution dramatically optimizes token usage; for instance, while a CLAUDE md dump might consume 22,000 tokens for 240 observations, Agentmemory achieves the same with only 1,900 tokens, representing a remarkable 92% reduction. Furthermore, as built-in memories become inefficient and invisible after just 1,000 observations, Agentmemory ensures 100% searchable and accessible memory retention. Benchmarked across 240 real coding sessions, it consistently delivers up to 95% fewer tokens per session, enables 200 times more tool calls before encountering context limits, and is proudly 100% open source, offering transparency and flexibility for advanced AI development.

Minions

Open source mission control for Hermes agent

While your Hermes Agent excels at single tasks, managing multiple agents in parallel can quickly descend into chaos, leading to silently failing cron jobs, blocked tasks, and excessive time spent on troubleshooting rather than achieving results. Minions provides a powerful, unified task board to oversee all your agent operations with clarity and control. Each running task benefits from periodic check-ins, automatic retries when stuck, and intelligent escalation only after genuinely exhausting all alternatives. Currently compatible with Hermes Agent, Minions is continuously expanding its support to include more runtimes, ensuring robust and reliable multi-agent management for your complex workflows.

Rosentic

Catch when coding agents break each other before merge

Rosentic offers a unique and critical pre-merge validation solution, meticulously checking every Pull Request (PR) against all other open PRs before merging. In environments where coding agents operate in parallel, they often introduce subtle conflicts that conventional single-PR analysis tools fail to detect. Rosentic is specifically designed to identify and resolve these complex inter-PR dependencies, ensuring code integrity. Our deterministic analysis guarantees consistent results every time, providing clear and actionable insights. Most scans return clean, giving you the confidence to merge swiftly. Rosentic runs securely on your existing infrastructure, requiring only one YAML file, no signup, and a rapid 60-second installation.

Gas City 1.0

build your own software factory

Gas City, the advanced successor to Gas Town, is an open-source software (OSS) platform engineered for software engineers to construct sophisticated software factories. This powerful platform enables the complete lifecycle management of software products, from building and deployment to operation and maintenance. Gas City allows you to seamlessly orchestrate your preferred Command Line Interface (CLI) coding agents, such as Claude Code, Codex, Gemini, and others, transforming their non-deterministic outputs into reliable, product-quality solutions. With Gas City, you gain unprecedented control over features, schedule, and quality, empowering you to truly pick three.

Re_gent

Version Control for AI agent Activity

re_gent acts as a robust version control system, akin to Git, specifically for your AI agent's actions. It provides unparalleled control, allowing you to undo, meticulously trace, and manage every single step your agent takes. With re_gent, you can clearly see what changes your coding agent made, identify the exact prompt that initiated those changes, and effortlessly roll back agent work across multiple files and even entire sessions, ensuring complete oversight and error recovery.

Ghost

Open-source, self-hosted game servers

Ghost-8 is an innovative open-source platform designed for effortlessly deploying and managing dedicated game servers. Instantly provision servers for popular titles such as Minecraft, Valheim, Rust, Palworld, Enshrouded, and Terraria directly within your existing Hetzner Cloud account. This solution offers unparalleled speed and control, allowing gamers and developers to establish high-performance gaming environments in mere seconds, without complex setup procedures.

WUPHF creates a collaborative virtual office powered by AI employees who autonomously build and maintain their own comprehensive knowledge base, ensuring they never lose context for any task you assign. This versatile platform supports a wide array of advanced agents, including Claude Code, Codex, and OpenClaw, along with local Large Language Models (LLMs) via OpenCode. Engage with your AI agents seamlessly through a Terminal User Interface (TUI), Web interface, or Telegram. WUPHF is open source, allowing you to run it securely on your own machine with your personal API keys, giving you full control and privacy over your operations.

PandaProbe

open source agent engineering platform

PandaProbe is an advanced open-source agent engineering platform engineered to provide unparalleled, deep observability into your AI agent applications. It empowers developers and operations teams with critical tools to trace execution paths, thoroughly evaluate performance, continuously monitor behavior, and efficiently debug AI agents across both development and production environments. By offering a comprehensive view of agent interactions and decision-making processes, PandaProbe ensures the reliability, predictability, and optimal performance of your intelligent systems, facilitating robust and scalable AI deployments.