AI Agent Fleet Dashboard

You have 50 agents running across three clouds, four frameworks, and a dozen services. Can you tell me which ones are healthy right now? Which one burned $200 overnight? Which one is stuck in a retry loop?

You can't. Because there's no single place to see them all.

AXME gives you a real-time fleet dashboard at mesh.axme.ai.

The Problem

AI agents are everywhere now. Your org runs them on different clouds (AWS, GCP, Azure), built with different frameworks (LangGraph, CrewAI, AutoGen, OpenAI Agents SDK), deployed as different services (Cloud Run, Lambda, ECS, bare VMs).

Each one has its own logs. Its own metrics. Its own way of telling you it's alive -- or not.

What you actually need:

• One screen showing every agent, every cloud, every framework
• Health status -- which agents are running, which are stuck, which crashed 10 minutes ago
• Cost tracking -- per-agent LLM spend, token counts, cost trends
• Policy enforcement -- rate limits, spending caps, kill switches
• Search and filter -- find agents by name, team, status, framework, cloud

What you have today: 12 browser tabs, 4 CLI tools, and a spreadsheet someone updates manually on Fridays.

The Solution

AXME Agent Mesh gives you a unified dashboard for your entire agent fleet. Every agent registers with a heartbeat. The dashboard shows them all in real time.

+--------------------------------------------------------+
|  AXME Agent Mesh Dashboard         mesh.axme.ai        |
+--------------------------------------------------------+
| Agents: 52 total | 48 healthy | 3 degraded | 1 dead    |
| Cost today: $142.30 | MTD: $3,847.12                   |
+--------------------------------------------------------+
| Name              | Status  | Cloud | Cost/hr | Uptime |
|-------------------|---------|-------|---------|--------|
| data-pipeline-01  | healthy | GCP   | $2.40   | 14d    |
| support-bot-prod  | healthy | AWS   | $8.10   | 7d     |
| code-reviewer     | degraded| GCP   | $0.90   | 2d     |
| invoice-processor | dead    | Azure | $0.00   | 0m     |
| research-agent-03 | healthy | GCP   | $1.20   | 5d     |
| ...               |         |       |         |        |
+--------------------------------------------------------+
| [Search] [Filter: status] [Filter: cloud] [Kill Agent] |
+--------------------------------------------------------+

What the dashboard gives you

• Agents table -- all registered agents with name, status, cloud, framework, team, cost, uptime
• Real-time health -- heartbeat-based status (healthy / degraded / dead), updated every 30 seconds
• Filters and search -- filter by status, cloud, framework, team; full-text search by agent name
• Cost breakdown -- per-agent LLM spend (tokens in/out, model, cost), hourly and daily aggregation
• Kill switch -- select an agent, click Kill, it receives a shutdown intent via AXME
• Policy view -- rate limits, spending caps, and escalation rules per agent or team

Quick Start

1. Install the SDK and CLI

pip install axme
# or: npm install @axme/sdk

2. Register your agent with a heartbeat

from axme import AxmeClient, AxmeClientConfig

client = AxmeClient(AxmeClientConfig(api_key=os.environ["AXME_API_KEY"]))

# Register this agent in the mesh
client.register_agent({
    "agent_id": "data-pipeline-01",
    "agent_type": "data_processor",
    "framework": "langgraph",
    "cloud": "gcp",
    "team": "data-eng",
    "metadata": {
        "region": "us-central1",
        "model": "claude-sonnet-4-20250514",
    },
})

# Start heartbeat (reports health + cost every 30s)
client.start_heartbeat(interval_seconds=30)

# Your agent does its work...
for task in task_queue:
    result = process(task)
    # Cost is tracked automatically via SDK instrumentation

// TypeScript
import { AxmeClient } from "@axme/sdk";

const client = new AxmeClient({ apiKey: process.env.AXME_API_KEY });

await client.registerAgent({
  agentId: "support-bot-prod",
  agentType: "customer_support",
  framework: "openai-agents",
  cloud: "aws",
  team: "support",
});

await client.startHeartbeat({ intervalSeconds: 30 });

3. Open the dashboard

# Open the fleet dashboard in your browser
axme mesh dashboard

# Or go directly to:
# https://mesh.axme.ai

4. Open the dashboard

# Open dashboard in browser with automatic SSO
axme mesh dashboard

Kill, resume, set policies - all from the dashboard UI at mesh.axme.ai. Or use the Python/TypeScript/Go/Java/.NET SDK programmatically.

How It Works

1. Registration -- agents auto-register when they send their first heartbeat via the SDK
2. Heartbeat -- the SDK sends a heartbeat every 30 seconds with buffered metrics (intents, latency, cost)
3. Dashboard -- the web UI at mesh.axme.ai reads the mesh state and renders the fleet view
4. Kill signal -- when you click Kill in the dashboard, the gateway blocks all intents to/from that agent
5. Policy enforcement -- cost limits and action allowlists are checked on every new intent; breaches return HTTP 429 or 403

Agent Lifecycle States

State	Meaning	Heartbeat
registering	Agent called register, first heartbeat pending	--
healthy	Heartbeat received within expected interval	On time
degraded	Heartbeat late (1-3x interval) or error reported	Late
dead	No heartbeat for 3+ intervals	Missing
killed	Shutdown intent sent and acknowledged	Stopped

Works With Any Framework

The dashboard doesn't care what framework your agents use. If they can call the AXME SDK, they appear in the dashboard.

Framework	Registration	Heartbeat	Kill
LangGraph	Yes	Yes	Yes
CrewAI	Yes	Yes	Yes
AutoGen	Yes	Yes	Yes
OpenAI Agents SDK	Yes	Yes	Yes
Google ADK	Yes	Yes	Yes
Pydantic AI	Yes	Yes	Yes
Raw Python/TS/Go	Yes	Yes	Yes

Links

• AXME -- agent coordination infrastructure
• AXME Docs -- full documentation
• AXME Cloud -- managed service
• Agent Mesh Dashboard -- live fleet dashboard

License

MIT