Skip to content

getting-started.md

Latest commit

 

History

History
131 lines (101 loc) · 3.36 KB

getting-started.md

File metadata and controls

131 lines (101 loc) · 3.36 KB
Raw

Getting Started

Prerequisites

  • Go 1.22+
  • An API key from at least one supported model provider

Installation

go get github.com/alanfokco/agentscope-go/pkg/agentscope

Your First Agent

package main

import (
    "context"
    "fmt"

    as "github.com/alanfokco/agentscope-go/pkg/agentscope"
    "github.com/alanfokco/agentscope-go/pkg/agentscope/agent"
    "github.com/alanfokco/agentscope-go/pkg/agentscope/model"
)

func main() {
    as.Init()

    cm, err := model.NewDashScopeChatModel(&model.DashScopeConfig{
        APIKey: "sk-...",
        Model:  "qwen-plus",
    })
    if err != nil {
        panic(err)
    }

    a := agent.NewUnifiedAgent("assistant", "You are a helpful AI assistant.", cm)

    reply, err := a.Reply(context.Background(), "Hello! What can you do?")
    if err != nil {
        panic(err)
    }

    if txt := reply.GetTextContent("\n"); txt != nil {
        fmt.Println(*txt)
    }
}

Adding Tools

Give your agent the ability to call functions:

weatherTool := tool.NewFunctionTool(
    "get_weather", "Get current weather for a city",
    json.RawMessage(`{
        "type": "object",
        "properties": {"city": {"type": "string"}},
        "required": ["city"]
    }`),
    func(ctx context.Context, input map[string]any) (any, error) {
        city, _ := input["city"].(string)
        return map[string]any{"city": city, "temp": "22°C"}, nil
    },
)

a := agent.NewUnifiedAgent("bot", "You are a weather assistant.", cm,
    agent.WithToolkit(tool.NewToolkit(weatherTool)),
    agent.WithReactConfig(agent.ReactConfig{MaxIters: 5}),
)

Streaming Responses

Get real-time output as the model generates:

ch, _ := a.ReplyStream(ctx, "Tell me a story.")
for evt := range ch {
    switch e := evt.(type) {
    case event.TextBlockDeltaEvent:
        fmt.Print(e.Delta)
    case event.ReplyEndEvent:
        fmt.Println()
    }
}

Environment Variables

Set one of these API keys to get started:

Variable Provider
DASHSCOPE_API_KEY Alibaba DashScope (Qwen)
ANTHROPIC_API_KEY Anthropic (Claude)
OPENAI_API_KEY OpenAI (GPT)
DEEPSEEK_API_KEY DeepSeek
GEMINI_API_KEY Google Gemini
MOONSHOT_API_KEY Moonshot (Kimi)
XAI_API_KEY xAI (Grok)

Optional: DASHSCOPE_BASE_URL to override the DashScope endpoint.

Running Examples

The project includes 35 examples. Run any of them:

export DASHSCOPE_API_KEY=sk-...
go run ./examples/agent_v2          # Tool calling
go run ./examples/streaming         # Streaming events
go run ./examples/multimodal        # Image input
go run ./examples/model_call        # Raw model API
go run ./examples/replay            # Deterministic replay
go run ./examples/pool              # Agent pool fan-out

See examples.md for the full list.

Next Steps

  • Architecture — Understand the package structure
  • Model Providers — Configure different LLM backends
  • Tools — Built-in tools, custom function tools, and document parsers
  • Middleware — Intercept and extend agent behavior (7 hooks)
  • Deployment — Run as an HTTP service, workspace sandboxing, agent pools
  • Go-Exclusive Features — Deterministic replay, fan-out pool, hot-reload, WASM sandbox, TCP mesh, load testing