Build your own
Cursor
with Zypher

A few lines of code to create powerful AI agents. Connect any MCP server, choose your LLM provider, and start building.

JSR
deno add jsr:@zypher/agent
/main.ts
import { createZypherAgent } from "@zypher/agent";
import { createFileSystemTools } from "@zypher/agent/tools";
import { eachValueFrom } from "rxjs-for-await";

const agent = await createZypherAgent({
  model: "claude-sonnet-4-5-20250929", // Or use ModelProvider for advanced options
  tools: [...createFileSystemTools()],
  mcpServers: ["@modelcontextprotocol/sequentialthinking-server"],
});

// Run task with streaming
const taskEvents = agent.runTask("Implement authentication middleware");

for await (const event of eachValueFrom(taskEvents)) {
  console.log(event);
}

Everything you need to build AI agents

A minimal yet powerful framework for creating AI agents with full control over tools, providers, and execution flow.

  • LLMInputOutputLearn & Adapt
    True Agent, Not Workflow
    A reactive execution loop where the agent decides each next step via LLM reasoning — no predefined workflow paths required.
  • Zypher
    Model & Provider Agnostic
    Works with Anthropic Claude, OpenAI GPT, and more models through a unified interface — with native support for each provider's unique features.
  • Task InputPlannerAgentResearchAgentCode AgentAgentQA AgentAgent
    Multi-Agent Architecture
    Break complex tasks into subtasks handled by specialized sub-agents — with automatic context sharing and coordinated handoffs.
  • ToolToken
    Extensible Interceptor Pipeline
    A post-reasoning pipeline with built-in tool execution, error handling, and token controls, plus support for custom interceptors.
  • MCPTools
    Rich Tools + MCP Protocol
    A comprehensive tool system for file ops, execution, vision, and documents, with MCP protocol support for unlimited extensibility.
  • Token EfficientMinimal Token LoadTraditionalHigh Token Usage
    Token-Efficient by Design
    Load only what's needed into context. Load agent skills and MCP tool definitions on-demand, programmatic tool use that keeps intermediate results out of context, and prompt caching — minimal tokens, maximum efficiency.

Ready to build your first agent?