DeepAgentLabs

HomeProductsControl Tower

AgenticOps Control Tower · Operate

Every agent,
one place.

One place to find, install and operate the DeepAgentLabs agent-operations stack. See which agents you run, how healthy they are and which tools each one has, then find the right tool for the next problem.

Open Control Tower View on GitHub
pip install agenticops-control-tower
  • Free and open source
  • Self-hosted: data stays on your machine
  • v0.2.0 on PyPI · pre-alpha
  • Built on the AI Operations Spec
FLEET OVERVIEW · EXAMPLELive
48registered agents
  • Healthy 41
  • Degraded 5
  • Unhealthy 2
  • Unknown 0
  • support-triageLangGraphhealthy
  • invoice-agentCrewAIdegraded
  • research-botOpenAI Agents SDKhealthy
  • ops-runnerAutoGenno heartbeat

Recent activity

  1. support-triage sent a heartbeat
  2. invoice-agent reported a slow tool call
  3. ops-runner stopped checking in

Example data, simulated live. Open Control Tower to see your own fleet.

Why it exists

More agents, less visibility.

Teams start with one agent and soon run dozens, across different frameworks, machines and owners. Without one shared list, nobody can answer simple questions quickly.

“Which agents do we even have?”

Agents live in different repos, notebooks and services. There is no single inventory.

“Is anything broken right now?”

An agent that silently stops working is often noticed only when a user complains.

“Which tools does each one use?”

Versions drift, and nobody knows which agent has AgenticLens, Sidecar or MCP, or which one is missing them.

Control Tower answers all three from one console: a fleet view of your agents, their health and the tools each one runs.

What it does

See every agent, know what needs attention.

Control Tower turns a scattered set of agents into one fleet you can read at a glance: who runs what, how healthy it is, and which tool to add next.

Fleet · Agentssample data
Agents table with filters for status, environment and capability

Fleet registry

Every agent Control Tower knows about, with its environment, where it runs, its framework, its tools and when it was last seen. Filter by status, environment or capability.

Fleet · Healthsample data
Health bar: healthy 1, degraded 1, unhealthy 0, unknown 0

Health status

One status per agent: healthy, degraded, unhealthy or unknown. A health bar across the fleet shows straight away where to look first.

Fleet · Capability coveragesample data
Capability coverage table: versions in use, coverage and missing agents per tool

Capability coverage

Which DeepAgentLabs tools each agent reports, the versions in use, and which agents are missing a tool, so gaps are easy to close.

Chat · Toolscatalog
Tools in the catalog with version and stage: AgenticLens, Agentic Chaos, Agentic Sidecar, Agentic Evals, Agentic MCP, Control Tower

Tool finder & catalog

Ask which tool fits a problem, or type @ to pick one. Browse tools, packages and use cases with their version, stage and install steps.

Two ways to run it

Your machine, or ours.

Both use the same console. Pick the setup that fits you.

Self-hosted

Available as a Python package

Run it yourself

Install Control Tower on your laptop or your own server. You sign in with an account that lives on that install, and nothing is sent to DeepAgentLabs: your agent inventory and history stay on the machine that runs it.

Account
A local account on your install. No DeepAgentLabs sign-up.
Where it runs
Your machine or your own server.
Where data lives
Local to that machine. No central database.
Install
pip install agenticops-control-tower

Best for: individual developers, and teams that need data to stay on their own infrastructure.

Hosted

Planned · preview only

Let DeepAgentLabs run it

DeepAgentLabs runs Control Tower in the cloud. You sign in from any browser and your workspace is stored centrally, so your whole team sees the same fleet without running servers.

Account
A DeepAgentLabs account.
Where it runs
DeepAgentLabs cloud.
Where data lives
Hosted database, per account.
Install
Nothing to install.

Best for: teams who want a shared workspace without installing or operating anything.

Inside the console

See it before you install it.

The Control Tower console runs in your browser. These screens are from the current prototype, with sample data.

Control Tower console, Chat: 'Find the right tool for your agents' with suggested questions, and the tools in the catalog with their versions on the right

Screens change every few seconds · click one to enlarge it

  • Chat
  • Fleet
  • Tools
  • Packages
  • Use cases
  • Guides

How it works

Up and running in four steps.

Steps 3 and 4 use commands that are in the repository and ship in the next release (they need the api extra).

  1. 1

    Install

    Add the Python package. Today it gives you the registry, discovery and status Python API.

    pip install agenticops-control-tower
  2. 2

    Describe your fleet

    Register your agents with where they run and which tools they use, or start from the example snapshot in the repository.

    examples/sample_fleet_snapshot.json
  3. 3

    Start the server

    Serve the console and API from your machine, loading a fleet snapshot.

    agenticops-control-tower serve --snapshot examples/sample_fleet_snapshot.json
  4. 4

    Check the fleet

    Open the Fleet page in the console, or ask for the fleet status from the command line.

    agenticops-control-tower status

Health at a glance

Four states, no guesswork.

Every agent gets one simple status, so the fleet view shows straight away where to look first.

  • HealthyChecking in and reporting no problems.
  • DegradedStill running, but reporting problems. Worth a look.
  • UnhealthyNot working as it should. Needs attention now.
  • UnknownNo status reported yet.

Part of the ecosystem

The operate step of the loop.

Every DeepAgentLabs tool speaks the same AI Operations Specification. Control Tower sits at the end of the loop, knows the whole catalog and shows which agent runs what.

  1. DefineAI Operations Spec
  2. ObserveAgenticLens
  3. StressAgentic Chaos
  4. EvaluateAgentic Evals
  5. OperateControl Tower

The stack in the Control Tower catalog

ToolWhat it's forVersionStage
AgenticLensTraces, cost and latencyv0.5.0Alpha
Agentic ChaosFault injection and drift testingv0.4.0Alpha
Agentic SidecarAllow, warn or block actionsv0.6.0Pre-alpha
Agentic EvalsScoring and CI gatesv0.7.0Beta
Agentic MCPEvery tool through one MCP interfacev0.3.0Alpha
Control TowerFleet registry and consolev0.2.0Pre-alpha

Versions as listed in the Control Tower catalog (updated 10 Oct 2026).

Works with any framework

  • LangGraph
  • CrewAI
  • OpenAI Agents SDK
  • AutoGen
  • LlamaIndex
  • Your own agents

Status

Where it is today.

Control Tower is pre-alpha and built in the open. Here's exactly what you can use now and what's coming.

Available now

  • v0.2.0 on PyPI
  • Registry, discovery and status Python API

Next release

  • serve command
  • status CLI
  • Both need the api extra

Preview

  • The console: Chat, Fleet, catalog and guides
  • Local sign-in

Planned

  • Hosted Control Tower
  • Console packaging
  • Fleet page connected to a running server
  • An AI model behind Chat

FAQ

Common questions.

Quick answers about cost, data, the chat and what works today.

Still have questions? Ask the community on Discord, or open an issue on GitHub. Feature ideas are welcome too.
Ask on Discord Open an issue
Is Control Tower free?

Yes. It's open source and free to use, like every DeepAgentLabs package.

Does my data leave my machine?

Not when you self-host. Your agent inventory and history stay on the machine that runs Control Tower, with a local account and no central database. The hosted version, when it arrives, stores your workspace in the DeepAgentLabs cloud.

Is the chat an AI model?

Not yet. Answers come from the DeepAgentLabs catalog of tools. Connecting an AI model is planned.

Can I use it today?

Yes, as a Python package: v0.2.0 on PyPI gives you the registry, discovery and status API. The serve command and the CLI ship in the next release, and the console is a preview.

Which frameworks does it support?

Any. The sample fleet includes agents built with LangGraph and the OpenAI Agents SDK, running on AWS Lambda and Docker.

Where do I ask questions or report bugs?

Open an issue on GitHub, request a feature from inside the console, or ask on Discord.

Try it

Open the Control Tower console.

Explore the console preview, or install the package and run it on your own machine.

Open Control Tower PyPI Ask on Discord