by israriqbal
Ultimate Multi-Agent OS for Autonomous AI NPCs 2026
# Add to your Claude Code skills
git clone https://github.com/israriqbal/agent-ecologiesGuides for using ai agents skills like agent-ecologies.
agent-ecologies is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by israriqbal. Ultimate Multi-Agent OS for Autonomous AI NPCs 2026. It has 73 GitHub stars.
agent-ecologies's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/israriqbal/agent-ecologies" and add it to your Claude Code skills directory (see the Installation section above).
agent-ecologies is primarily written in HTML. It is open-source under israriqbal on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh agent-ecologies against similar tools.
No comments yet. Be the first to share your thoughts!
Unlocks once the catalog security scan passes (runs nightly).
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
Where AI Characters Come Alive: A Living, Breathing Digital Ecosystem for Autonomous NPCs
Imagine a pixel-art world where artificial intelligence agents don't just respond—they live. They wake up, wander through sun-dappled forests, strike up conversations with strangers, learn new skills, and form relationships. Welcome to Agent-Worlds, the multi-agent operating system designed for LLM-powered non-player characters that exist autonomously in a persistent 2D environment.
This is not a chatbot bolted onto a game. This is a sentient microcosm where every character has memory, purpose, and the freedom to evolve. Built on the Model Context Protocol (MCP), Agent-Worlds orchestrates interactions between multiple large language models—whether it's DeepSeek's raw reasoning, OpenAI's conversational fluency, or Claude's nuanced dialogue—to create characters that feel genuinely alive.
Agent-Worlds reimagines how AI characters inhabit digital spaces. Instead of scripted dialogue trees or pre-defined behavior loops, each agent operates as an independent cognitive entity with its own goals, knowledge base, and personality matrix. The world persists even when no human is watching—characters continue their daily routines, explore new territories, and engage in emergent social dynamics that no programmer could have predicted.
This platform serves as both a sandbox for AI researchers studying emergent behavior and a toolkit for game developers who want NPCs that surprise, delight, and evolve alongside players.
Agent-Worlds follows a modular, event-driven architecture where each component operates independently yet harmoniously.
World Kernel: The central runtime environment that manages time, physics, and spatial relationships. Every pixel in the 2D grid has properties—walkable terrain, resource nodes, collision boundaries—that agents perceive and interact with.
Agent Brain: Each NPC runs its own instance of a language model wrapper. The Brain processes sensory input (what the agent sees, hears, and remembers) and produces behavioral output (movement decisions, dialogue generation, skill execution). Multiple LLM backends can be assigned per agent, allowing hybrid intelligence.
Memory Vault: A dual-tier storage system. Short-term memory handles recent 50-100 interactions, while long-term memory compresses significant events into abstract summaries. Memories decay and reinforce based on emotional weight and repetition.
Skill Engine: Skills are JSON-defined capability modules that agents can invoke. From "woodcutting" to "negotiation," each skill has prerequisites, success rates, and experiential growth mechanics. The engine tracks proficiency and unlocks advanced variants over time.
Social Nexus: Manages character relationships using a multi-dimensional graph. Each relationship has affinity, trust, history, and role attributes. These influence how agents interact, share information, and form coalitions.
MCP Broker: The Model Context Protocol implementation that standardizes how agents send and receive context. This decouples the LLM selection from the application logic, enabling hot-swappable intelligence.
To bring your first agent world to life, you'll need to configure the environment and spawn your initial inhabitants.
Game Development: Create RPG towns where NPCs have their own schedules, gossip networks, and evolving economies. Players can build relationships with characters who remember past encounters.
AI Research: Study emergent social behaviors, information propagation, and skill transmission in controlled digital environments. Run experiments on group dynamics with reproducible parameters.
Story Generation: Generate emergent narratives by letting autonomous characters live their lives. The world writes its own drama through agent interactions.
Education & Training: Build simulations where students interact with AI characters designed to teach specific skills or knowledge domains, adapting their teaching style to learner behavior.
The skill system in Agent-Worlds transforms static NPCs into lifelong learners. Every skill belongs to one of five families:
Agents gain experience through successful use and lose proficiency through neglect. Skills unlock specializations at mastery thresholds:
The Social Nexus models relationships using a 3-dimensional vector: Affinity (-100 to 100), Trust (0 to 100), and Familiarity (0 to 100). These dimensions interact dynamically:
Relationships decay by 1-3% per simulated day without interaction and can trigger emotional events that broadcast to connected agents.
Agent-Worlds is designed to scale from tiny hamlets to sprawling metropolises.
| World Size | Agent Count | RAM Usage | LLM Calls/Hour (Typical) |
|---|---|---|---|
| Small (32x32) | 5-20 | 1-2 GB | 60-200 |
| Medium (64x64) | 20-100 | 4-8 GB | 200-1000 |
| Large (128x128) | 100-500 | 16-32 GB | 1000-5000 |
| Massive (256x256) | 500-2000+ | 64+ GB | 5000+ |
Performance optimizations include batched LLM calls, tiered memory compression, and spatial partitioning for efficient pathfinding.
Characters can be configured to communicate in any language supported by the underlying LLM. The world itself uses a language-agnostic encoding for internal states. Translation hooks allow agents to learn new languages as skills, enabling cross-cultural interactions within the same world.
The platform exposes several extension points:
Agent-Worlds supports a growing ecosystem of shared worlds, character archetypes, and skill packs contributed by developers worldwide. The MCP protocol ensures cross-compatibility between modules created by different authors.
All agent data remains local unless you explicitly enable cloud synchronization. API keys are stored in encrypted configuration files. The world kernel runs with sandboxed permissions, preventing agents from executing arbitrary system commands.