AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
翻譯準備中
Minimal agent runtime for any OpenAI-compatible LLM
翻譯準備中
取自儲存庫 README
正在載入完整 README。
檢視完整 README ↗go get github.com/atakang7/axon/v2
Axon uses gopkg.in/yaml.v3 for configuration and golang.org/x/sys/windows
for native Windows process jobs.
On Windows, shell tools use Git for Windows Bash when installed, with a PowerShell fallback. Background commands and foreground cancellation use Windows Job Objects to terminate the whole process tree. Windows does not provide POSIX SIGTERM, so job termination is immediate. Linux and macOS continue to use POSIX process groups.
Point it at any OpenAI-compatible endpoint and drive a turn:
package main
import (
"context"
"fmt"
"log"
"os"
"github.com/atakang7/axon/v2"
)
func main() {
model, err := axon.OpenAI(axon.ClientConfig{
Provider: axon.Provider{
BaseURL: "https://openrouter.ai/api",
Model: "deepseek/deepseek-v3.2",
APIKey: os.Getenv("OPENROUTER_API_KEY"),
},
})
if err != nil {
log.Fatal(err)
}
agent, err := axon.New(axon.Config{
Model: model,
SystemPrompt: "You are a careful coding agent.",
})
if err != nil {
log.Fatal(err)
}
defer agent.Close() // kills background shells, closes MCP
result, err := agent.Step(context.Background(), "List the files in this directory.")
if err != nil {
log.Fatal(err)
}
fmt.Println(result.Assistant)
}
Set an API key and run:
export OPENROUTER_API_KEY=sk-...
go run .
The snippet above is the full agent loop — one client, one agent, one Step. Axon layers the rest on top when you need it:
Full documentation lives at atakang7.github.io/axon:
Getting Started
Core Concepts
Step() drives the agent through model calls and tool execution.Tools
Configuration
Under the Hood