FrostAgent

An AI Agent orchestration + API middleware service built with Golang.

English | 中文

Notice

This project is still in early stages and is intended for personal research use. We welcome PRs and guidance from everyone!

Collaboration with ActionsCat

ActionsCat supports the automation of static orchestration workflows.

After integrating ActionsCat into the adapter, you can operate both simultaneously, and the agent will demonstrate its outstanding capabilities.

Quick Start

1. Build the Project

go build -o frostagent.exe

2. Configure Environment Variables

Create a .env file or set system environment variables:

# Upstream API endpoint (e.g., Alibaba Cloud Tongyi Qianwen)
set UPSTREAM_ENDPOINT=https://dashscope.aliyuncs.com/compatible-mode/v1

# Upstream API key
set UPSTREAM_API_KEY=sk-your-api-key-here

# Middleware listening address (default: :8080)
set LISTEN_ADDR=:8080

3. Start the Service

go run ./cmd/app

API Usage

Health Check

curl http://localhost:8080/health

Response:

{"status":"ok"}

Chat Completion Endpoint

curl -X POST http://localhost:8080/agent/query \
  -H "Content-Type: application/json" \
  -d '{
    "input": "Tell me about the weather in Beijing."
  }'

Multi-context / chat history is also supported. input is optional when messages is provided; if both are provided, input is appended as the latest user message.

curl -X POST http://localhost:8080/agent/query \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      {"role": "user", "content": "My name is Frost."},
      {"role": "assistant", "content": "Nice to meet you, Frost."},
      {"role": "user", "content": "What is my name?"}
    ]
  }'

Custom Upstream Service

FrostAgent can proxy to any OpenAI-compatible API endpoint. Simply modify the environment variables to switch upstream services.

Routes

Method	Endpoint	Description
GET	/health	Health check
POST	/agent/query	Chat completion endpoint

Environment Variables

Variable	Description	Default Value
UPSTREAM_ENDPOINT	Upstream API endpoint URL	https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions
UPSTREAM_API_KEY	Upstream API authentication key	sk-xxx
LISTEN_ADDR	Middleware service listening address	:8080
MAX_CONTEXT_MESSAGES	Max messages kept in context, including system prompt	20
MAX_CONTEXT_CHARS	Approximate max context characters before trimming	24000

License

MPL-2.0 (see LICENSE file)