Newly Accepted

Forked is an open-source system for agent state management, designed to fork, rewind, replay, and persist AI agent state with effectively-once side effects. It includes a Rust core, a self-hosted server, cloud deployment, a TypeScript SDK, LangGraph support, and an MCP server.

A macOS Markdown editor designed to work smoothly with AI agents editing the same document at the same time. It uses Y.js to merge concurrent changes, preserves Markdown formatting details, and includes a live preview with support for GitHub-flavored Markdown and mermaid diagrams.

Lumabri is an open-source distributed inference tool that runs large AI models across multiple computers and servers. It includes a terminal UI and browser interface, with setup, networking, memory budgeting, and calibration details documented in the README.

A Claude Code mod that prevents runaway subagent fan-out, heavy-context prompts, and retry loops from burning through usage limits too quickly.

Anthropic’s research post shows Claude proving security properties across 123 open-source libraries and discusses how LLMs can support formal verification workflows for real software.

Axminster is an AI-augmented software design tool for model-driven engineering. It supports UML 2.5.1 and OCL 2.4, can save models as XMI, and includes an MCP server for sharing designs with LLM-based coding assistants.

Open-source terminal coding agent in a single native binary, with steerable subagents, token-saving output filters, a workbench, file revert tools, and support for multiple providers or local models.

Self-hosted project management for agencies and client work, with tickets, time tracking, docs, meetings, and an AI assistant that cites its sources. The repo also exposes an MCP server and REST API for external agents and tools.

A research writeup from AgentSploit analyzing 1,384 tools across 28 MCP servers, highlighting how tool descriptions, source tools, and sink tools create an attack surface for agents.

Nines is an open-source SLO platform for defining critical user journeys as code, measuring reliability with Prometheus or Mimir, and managing workflows through a web UI, CLI, Go SDK, and MCP tools.

Airlock is a Mac app that runs Claude Code or Codex inside a virtual machine, keeping agent changes on a copy of your project until you review and apply them.

Self-hosted memory layer for coding agents. It stores notes as Markdown in a local folder and exposes them over MCP so tools like Claude Code, Codex, and Cursor can recall shared context.

A CLI and GitHub Action for testing MCP servers with checks for tool descriptions, safety labels, hidden instructions, bad inputs, secret leaks, OAuth setup, resources, prompts, and performance.

A Claude Code plugin marketplace with 42 skills for building, shipping, and announcing apps across App Store, Google Play, Mac, Windows, and web workflows.

A GitHub repository for a Claude Code plugin and helper that shows live per-step progress for long-running shell commands instead of a generic spinner.

A detailed build post about creating Live Coloring Book with AI coding agents, including technical details on animation rigging, photo-based crayon color reading, browser-side processing, and what went wrong during development.

Open-source MCP server for financial analysis that brings 500+ calculated metrics, valuation models, technical indicators, and macro data into Claude, ChatGPT, Cursor, Copilot, and other MCP-capable clients.

A small character-level AI model that generates realistic-sounding town names in the browser. It includes interactive controls, model-size comparisons, and technical details on training, data, and inference.

A native macOS workspace for coordinating coding agents, with workflows, cross-agent memory, traces, and support for several agent harnesses.

Skillstore is a searchable skills library for Claude Code and Codex that delivers task-specific instructions and supporting files through MCP. It also lets teams create, share, and improve their own methods with feedback.

An embedded Rust database that speaks SQLite and adds built-in vector and keyword search, plus CLI, shell, and MCP server support for agentic workflows.

A research project exploring multi-agent LLM coordination in Pokémon Showdown. It compares talk modes, reports win rates and token costs, and frames the environment as a benchmark for studying collaboration between agents.

Open-source managed agent platform that can be self-hosted or used via Omnara Cloud. The page shows a concrete quickstart, CLI flow, TypeScript SDK usage, and self-hosting instructions for launching AI agents with tools, permissions, streaming, and sandboxes.

Anthropic research post showing Claude applied to attack surface analysis and reduction across 25 open-source codebases, with concrete findings and security-focused methods.