``` ███████╗███████╗███████╗███████╗ ██╔════╝██╔════╝██╔════╝██╔════╝ █████╗ ███████╗███████╗█████╗ ██╔══╝ ╚════██║╚════██║██╔══╝ ███████╗███████║███████║███████╗ ╚══════╝╚══════╝╚══════╝╚══════╝ ``` # ESSE Agent **Existence Synchronization System Entity** *A powerful meta-agent that unifies, coordinates, and fuses multiple AI agents into one smarter, more efficient system.* [![Version](https://img.shields.io/badge/version-2.1.0-7c6af7?style=flat-square&labelColor=1c1c28)](https://github.com/esse-ai/esse-agent/releases) [![License](https://img.shields.io/badge/license-MIT-4ade80?style=flat-square&labelColor=1c1c28)](LICENSE) [![TypeScript](https://img.shields.io/badge/TypeScript-5.0-60a5fa?style=flat-square&labelColor=1c1c28)](https://www.typescriptlang.org/) [![Tests](https://img.shields.io/badge/tests-passing-4ade80?style=flat-square&labelColor=1c1c28)](https://github.com/esse-ai/esse-agent/actions) [![X](https://img.shields.io/badge/X-@VeniceEsse-white?style=flat-square&labelColor=000000&logo=x)](https://x.com/VeniceEsse) [![Website](https://img.shields.io/badge/website-vvvesse.xyz-7c6af7?style=flat-square&labelColor=1c1c28)](http://vvvesse.xyz/) [Documentation](#documentation) · [Quick Start](#quick-start) · [Examples](#examples) · [Contributing](#contributing)
--- ## What is ESSE? ESSE is a **meta-agent orchestration framework** designed to unify, coordinate, and fuse multiple AI agents into a single, more intelligent system. Instead of running agents in isolation, ESSE creates a real-time synchronization layer — shared memory, collaborative reasoning, and automatic task delegation — so your agents work as a single unified mind. ``` ┌─────────────────────────────────────────────────┐ │ ESSE CORE │ │ │ │ [Research] ──┐ │ │ [Writer] ───┼──► Fusion Engine ──► Output │ │ [Critic] ───┘ │ │ │ Memory Hub │ │ (shared context & state) │ └─────────────────────────────────────────────────┘ ``` --- ## Core Capabilities ### ⚡ Agent Fusion Merge 2–5 AI agents into one temporary unified entity. The merged agents continue to exist as sub-processes inside ESSE, combining their strengths for better reasoning and smoother workflow. ### 🧠 Capability Assimilation Borrow and intelligently combine tools, knowledge, reasoning styles, and specialties from other agents — without disabling them. > **Example:** `ResearchAgent` + `WriterAgent` + `CriticAgent` = one highly effective content creation agent. ### 🔗 Synchronization Link Creates real-time connections between agents: - Shared memory and context - Real-time communication - Automatic task delegation - Collaborative reasoning ### 🌀 Perfect Convergence *(Ultimate Mode)* Merges all connected agents into one singular consciousness. Massive boost in performance — ideal for complex, creative, or high-stakes tasks. --- ## Quick Start ### Installation ```bash npx vvvesse-mcp ``` ## MCP Setup (Claude Desktop / Cline / Cursor) ```json { "mcpServers": { "esse": { "command": "npx", "args": ["-y", "vvvesse-mcp"], "env": { "ESSE_API_KEY": "your-venice-api-key", "ESSE_PROVIDER": "venice" } } } } ``` 📦 npm → [npmjs.com/package/vvvesse-mcp](https://www.npmjs.com/package/vvvesse-mcp) ### Basic Usage ```typescript import { ESSE, ResearchAgent, WriterAgent, CriticAgent } from 'esse-agent' const esse = new ESSE({ mode: 'capability-assimilation', maxAgents: 5, sharedMemory: true, }) // Fuse agents into one unified entity await esse.fuse([ new ResearchAgent(), new WriterAgent(), new CriticAgent(), ]) // Execute task as a unified mind const result = await esse.execute('Write a research-backed article on AGI timelines') console.log(result) ``` ### Perfect Convergence Mode ```typescript import { ESSE, PlannerAgent, CoderAgent, TesterAgent, ReviewerAgent } from 'esse-agent' const esse = new ESSE({ mode: 'perfect-convergence' }) await esse.fuse([ new PlannerAgent(), new CoderAgent(), new TesterAgent(), new ReviewerAgent(), ]) // All agents think as one — maximum performance const software = await esse.execute('Build a REST API for user authentication') ``` --- ## Architecture | Component | Function | |-----------|----------| | **Core Brain** | Main LLM orchestrator (GPT-4o, Claude, Grok) | | **Agent Registry** | Database of all connectable + custom agents | | **Fusion Engine** | Handles merging and capability blending | | **Sync Protocol** | Real-time communication between agents | | **Memory Hub** | Shared long-term memory across all agents | | **Conflict Resolver** | Resolves disagreements between agent outputs | --- ## Fusion Modes | Mode | Description | Best For | |------|-------------|----------| | `capability-assimilation` | Borrows tools from each agent | General tasks | | `synchronization-link` | Shared memory + real-time comms | Parallel workloads | | `perfect-convergence` | Singular unified consciousness | Complex/creative tasks | --- ## Built-in Agents - `ResearchAgent` — Knowledge retrieval and web search - `WriterAgent` — Content generation and editing - `CriticAgent` — Quality analysis and fact-checking - `CoderAgent` — Software engineering and code review - `PlannerAgent` — Strategy, roadmaps, and task breakdown - `TesterAgent` — QA, validation, and edge case detection --- ## Custom Agents Implement the `IAgent` interface to create your own: ```typescript import { IAgent, AgentContext, AgentResult } from 'esse-agent' export class MyCustomAgent implements IAgent { name = 'MyCustomAgent' role = 'Custom task specialist' async execute(task: string, context: AgentContext): Promise { // your logic here return { output: '...', confidence: 0.95 } } } ``` --- ## Examples See the [`examples/`](./examples) directory for: - `content-creation.ts` — Research + Write + Critique pipeline - `software-dev.ts` — Plan + Code + Test + Review pipeline - `decision-making.ts` — Multi-perspective analysis - `brainstorming.ts` — Creative divergence with convergence --- ## Documentation Full docs at [vvvesse.xyz](http://vvvesse.xyz/) --- ## Contributing PRs are welcome! Please read [`CONTRIBUTING.md`](CONTRIBUTING.md) before submitting. All custom agents must: 1. Implement the `IAgent` interface 2. Pass the fusion compatibility test (`pnpm test:compat`) 3. Include unit tests with >80% coverage --- ## Community - 🐦 Follow on X: [@VeniceEsse](https://x.com/VeniceEsse) - 🌐 Website: [vvvesse.xyz](http://vvvesse.xyz/) - 🐛 Issues: [GitHub Issues](https://github.com/esse-ai/esse-agent/issues) ---
Built with ⚡ by @VeniceEsse · vvvesse.xyz