AXME
AXME is a managed durable execution platform for AI agents and services, providing workflow orchestration with built-in human approval gates, crash recovery, and multi-agent coordination without requiring cluster setup. It solves the problems of fragile polling/webhook glue code, lost state on agent crashes, and lack of human-in-the-loop support in existing orchestration tools. It is for backend teams, AI agent builders, and platform teams seeking a simpler alternative to Temporal.
✨ Key features
- Built-in human approval gates with 8 task types
- AI agent crash recovery preserving intent state
- Multi-agent orchestration across machines and languages
- Managed service, no cluster setup required
- Agent monitoring dashboard and kill switch
- SDKs for Python, TypeScript, Go, Java, .NET
🎯 Use cases
- Orchestrate long-running approval workflows with human-in-the-loop steps
- Coordinate multiple AI agents across services and clouds
- Replace polling and webhook glue code with durable intent delivery
- Recover AI agent work after crashes without restarting from scratch
- Monitor and control production agents with health and cost tracking
📦 Installation
🧰 Requirements: Requires an AXME account and API key (axme_sa_...). CLI installation via curl script. SDKs available for Python, TypeScript, Go, Java, .NET.
curl -fsSL https://raw.githubusercontent.com/AxmeAI/axme-cli/main/install.sh | shaxme loginaxme examples run human/cliPython SDK: pip install axme
TypeScript SDK: npm install @axme/axme
Go SDK: go get github.com/AxmeAI/axme-sdk-go@latest
Java SDK: Maven Central ai.axme:axme-sdk
.NET SDK: dotnet add package Axme.Sdk
🚀 Usage
from axme import AxmeClient, AxmeClientConfig
client = AxmeClient(AxmeClientConfig(api_key="axme_sa_..."))
intent = client.send_intent({
"intent_type": "intent.deployment.approval.v1",
"to_agent": "agent://myorg/prod/deploy-checker",
"payload": {"service": "api-gateway", "version": "3.2.1"},
})
result = client.wait_for(intent["id"]) # waits hours if needed. survives crashes.
⚠️ Good to know
All SDKs are at v0.1.2 (Alpha), indicating early-stage software with potential instability.
❓ FAQ
How does AXME handle human approvals?
AXME provides built-in human-in-the-loop task types (approval, review, form, etc.) that wait as long as needed, with participation via CLI, email magic links, or forms.
What happens if an agent crashes mid-task?
AXME preserves intent state across restarts, so work resumes from where it left off rather than starting over.
Do I need to set up a cluster to use AXME?
No, AXME is a managed service, so no cluster setup is required; you just use the SDKs or CLI.
Can AXME orchestrate multiple agents across different machines?
Yes, AXME provides an agent coordination protocol that works across services, languages, and clouds.
📊 Repository
🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.