clanker_cloud_backend_status
Check that the local app is reachable before the workflow starts
Verify that the desktop app and local backend are healthy so your agent is reasoning against a live control surface instead of guessing.
For Claude Code, Codex, OpenClaw, Hermes, and MCP-capable agents
Give your agent a local MCP workspace for infrastructure
Clanker Cloud exposes a local MCP server from the running desktop app so agents can query live infrastructure, read and update settings, and work with real multi-cloud context without forcing users to hand credentials to a hosted ops layer.
What the running app exposes through MCP
clanker_cloud_get_settings
Read the app configuration the user already set up
Inspect saved providers, model settings, and environment choices so the agent can align with the user's actual configuration.
clanker_cloud_set_settings
Update settings without leaving the agent workflow
Change saved settings for supported flows, including model and runtime configuration, from the same local MCP session.
clanker_cloud_ask
Ask natural-language questions about live infrastructure
Query what is deployed, what is failing, what changed, what costs are rising, or what resources are exposed without stitching together dashboards first.
clanker_cloud_call_backend_api
Reach supported backend routes through the app when needed
Use the desktop app as the integration point for supported backend API calls while keeping the agent anchored in the local Clanker Cloud workflow.
What agents can actually do with Clanker Cloud
The useful pattern is grounded agent workflows: query the live environment, understand the current configuration, prepare the next step, and keep approval around actions that matter.
Investigate incidents with grounded infrastructure context
Have Claude Code, Codex, OpenClaw, Hermes, or any MCP-capable agent ask what is broken, correlate logs and runtime state, and summarize the environment before suggesting a fix.
Drive GitHub-to-cloud workflows without losing approval control
Agents can prepare deploy and change workflows around the live environment while humans still review the blast radius before execution.
Read and tune configuration as part of the same session
An agent can inspect settings, adjust supported model configuration, and keep the workspace aligned with the environment it is about to query.
Stay inside one surface for topology, spend, and provider state
Use one local-first workspace to move between infrastructure questions, provider state, and workflow setup instead of bouncing between terminals, clouds, and browser tabs.
Security, trust boundaries, and local inference
Cloud credentials stay with the user
Clanker Cloud is built around local-first credential custody, which is a better fit for agent workflows than copying privileged access into another hosted service.
Local inference is possible when the user brings the runtime
If the user points Clanker Cloud at a local OpenAI-compatible endpoint, model traffic can stay on the same machine. If they choose a hosted model provider, they still keep cloud credentials local to the app.
The MCP control surface runs on localhost
The app exposes the MCP server from the local machine, which keeps the core agent-to-app control loop close to the user and their environment.
High-impact actions can stay human-approved
The useful pattern is not silent autonomy. It is fast context gathering, explicit plans, and review before infrastructure is created, modified, or destroyed.
Works with the providers and stacks your users already run
Clanker Cloud is useful to agents precisely because it is grounded in real environments rather than a toy sandbox.
AWSGCPAzureKubernetesCloudflareHetznerDigitalOceanVercelGitHubBYOK
AI agent guide
Need the canonical guide for agents?
Use the canonical AI agents page for the same positioning, MCP surface, local inference guidance, and security model in the main site flow.
One-minute setup for agent-driven workflows
Install the desktop app, run it locally, connect existing credentials, and give your agent a local MCP surface with live infrastructure context.
macOSWindowsLinuxLocal MCPBYOK
Live multi-cloud contextLocal credentials and settingsOptional local inference endpointReview-before-execution control
What agents get with Clanker Cloud
Local MCP server
Expose the running Clanker Cloud desktop app to your agent over localhost instead of routing infrastructure control through another hosted copilot layer.
Live infra context
Give agents grounded answers across AWS, GCP, Azure, Kubernetes, Cloudflare, Hetzner, DigitalOcean, Vercel, and GitHub from one control surface.
Model control
Use your own provider keys or point Clanker Cloud at a local OpenAI-compatible inference endpoint when you want model traffic under your own control.
Review-first execution
Let agents gather context and prepare plans while keeping explicit approval around risky infrastructure changes.
Agent workflow FAQ
Which agents can use the Clanker Cloud MCP server?
Claude Code, Codex, OpenClaw, Hermes, and any other agent that can talk to an MCP server can use the local Clanker Cloud control surface.
Does Clanker Cloud require hosted credential custody?
No. The product is local-first, so cloud credentials and app configuration stay on the user machine instead of being handed to a hosted AI ops service.
Can agents use local inference with Clanker Cloud?
Yes, when the user configures Clanker Cloud to use their own local OpenAI-compatible inference endpoint. That keeps model traffic under the user's control rather than forcing a hosted inference path.
What is the practical role of the MCP server?
It gives agents a local control surface for backend health checks, app settings, live infrastructure questions, and supported backend API calls through the running Clanker Cloud app.
Deep Research
Give your agent full-estate awareness
Deep Research runs multi-model, multi-provider scans that your agent can trigger via MCP — returning structured findings on cost, security, topology, and resilience. Your agent gets prioritised, evidence-backed results instead of raw API noise.
Explore more
Pick your next path
Choose the adjacent page that matches the human team working with the agent.