M

multi-agent

Toolkit local-first de agentes IA para OpenCode. Orquesta subagentes cloud (OpenCode Go) y locales (Ollama), con memoria persistente (engram), OCR local (uocr) y generación/edición de imágenes (ComfyUI). Incluye dos agentes primary — agente-pobre (orquestador, optimización de costo) y agente-hacker-pobre (pentesting ético autorizado + auditoría de seguridad) — con skills de SAST/pentest, y un vault Obsidian de conocimiento. TypeScript (bun) + plugins de OpenCode.

Multi-platform AI dev-team orchestration with a single canonical knowledge base. Plan-Approve-Execute workflows, validation gates, projected to GitHub Copilot, Gemini CLI, Opencode, and Cursor.

Contains the fundamental classes and interfaces that define the basic functionality of the SCOP framework.

Desktop kanban board for the markdown tickets your AI coding agents leave behind. Reads files directly — Claude, Codex, Gemini, Copilot, local models all push to the same folder.

A framework for working with multiple overlapping AI coding agents: one shared board, one human who makes the decisions.

MCP server + harness for the Linux edition of Sid Meiers Civilization V: Brave New World. Allows agentic models to see and play the game the same way a human would - fog of war, diplomacy, pesky barbarians. Does not enable model to control background AIs.

Open specification of cognitive-agent safety: a 27-state geometric model (Z3^3), 16 AoC defenses (11 context-free + 5 trajectory), Ed25519 agent identity, 10 Agda proofs with 0 postulates, 33-requirement conformance suite. Apache-2.0 reference implementation.

One memory, one set of guardrails and one MCP hub for all your coding agents -- local-first. Read-only mirror of https://github.com/Nokido-labs/nokido : issues and contributions there.

Empirical validation of C4 geometric defense against 16 Agents of Chaos. 550 adversarial prompts. 4 defense systems. 96.7% block rate. LLM validation on GPT-4o-mini + Mistral 7B. MIT.

Multi-agent reinforcement learning testbed: a top-down 2D defense game where a team of agents must protect a flag from waves of enemies that get faster and more numerous over time. Fully batched PyTorch simulation on CUDA (thousands of arenas in parallel), PPO / MAPPO with MLP, GRU, LSTM and attention policies, benchmarks. The goal: cooperation that emerges from a shared reward.

Self-hosted multi-agent research analyst, in progress: ask a question, AI agents research the web and your own documents, and you get a cited report. FastAPI, Postgres/pgvector, Redis, Next.js, Docker Compose.

Hunt, Verify, Report. Scope-locked agent fleet on GitLab Duo Agent Platform and the a2a-omega mesh. Path B: Bring Your Own. Hands-off.

Deterministic state machine and multi-agent narrative execution pipeline with GitOps cascades, RAG vector memory, and asynchronous asset synthesis.

Configurable, review-only Agent Skill that coordinates specialized reviewers across local changes, Git refs, GitHub pull requests, and GitLab merge requests.

MCP-сервер обміну між двома агентами над одним репозиторієм (Rust + SQLite). Канон: https://github.com/RiasJ1Dar/agent-exchange

Bounded, receipt-producing task handoffs between independently authenticated local agent harnesses.

CivicOS AI is an open-source AGPL-3.0+ AI operating system designed to unify automation, coordination, and intelligence across public and private systems. It provides a modular runtime for agents, workflows, and plugins, enabling organizations to build, deploy, and interconnect autonomous business and civic processes. Through its interoperable architecture, CivicOS AI turns complex operations into composable, auditable, and scalable intelligent systems. https://roxanneardary.com/civicos-ai/

ChangeCore is a content intelligence system that continuously improves web pages by detecting outdated information, tracking relevance shifts, and aligning content with real-world truth and evolving user intent. It combines multi-agent analysis, external data ingestion, and AI search visibility optimization to ensure content remains accurate, discoverable, and useful over time. https://roxanneardary.com/changecore/

SharedMind is a collective intelligence system that listens to humanity, identifies important problems, connects knowledge and experiences, and uses multi-agent analysis to uncover existing and new possibilities for solving what matters. https://roxanneardary.com/sharedmind/