ClawMetry Agent Builder
Guides

Agent building guides

Fourteen runtimes you can deploy to a private micro-VM in a couple of minutes. These guides cover what each one is actually good at, which chat channels it speaks, which key it needs, and what it costs to keep running — so you pick once and pick right.

Start from the job

You know what you want the agent to do. These arrive at the runtime.

Guide Build a Telegram assistant that never sleeps A step-by-step guide to putting an AI agent behind a Telegram bot so it answers while you are asleep — which runtime to pick, and what it costs per month. Guide Build a WhatsApp agent WhatsApp is the hardest channel to self-host well. Here is which runtime speaks it, how the pairing works, and how to keep the session alive. Guide Build an always-on code reviewer Put a coding agent on its own micro-VM with a clone of your repo, and review pull requests by sending it a message. Which runtime, which key, what it costs. Guide Build an agent that runs on a schedule Morning digests, hourly monitors, weekly reports. Which runtimes have real cron, how scheduled tasks differ from chat, and how to stop them running away. Guide What an always-on agent really costs Hosting and model tokens are two different bills and only one is predictable. A worked breakdown of both, and the failure modes that make the second spike.

Start from the runtime

Five runtimes with genuinely different architectures. Read these if you already know roughly what you are building.

OpenClaw · Plus $19/mo Build an OpenClaw agent Full-featured personal agent gateway: the kitchen sink NanoClaw · Plus $19/mo Build a NanoClaw agent Secure agent that runs each Claude agent in its own container PicoClaw · Standard $9/mo Build a PicoClaw agent Ultra-light Go agent: 16 chat channels in a <10 MB binary Hermes · Standard $9/mo Build a Hermes agent Multi-agent workflow orchestration NemoClaw · Secure Box $59/mo Build a NemoClaw agent NVIDIA's multi-agent box: several sandboxed OpenClaw or Hermes agents on one dedicated VM

Hosted coding agents

Nine coding agents in the same shape: a full agent in an isolated micro-VM with a persistent workspace, reachable from chat. They differ in the harness, the key, and the temperament — compared side by side here.

Standard · $9/mo Host Claude Code The reference implementation. If you have no strong opinion about which coding agent to host, host this one — it has the deepest tool use, and it is the only runtime here that can run on a Claude subscription instead of metered API tokens. Standard · $9/mo Host Codex Pick this when your team is already standardised on OpenAI models and you want the coding agent to match. Same shape as hosted Claude Code — persistent workspace, Telegram front door — with OpenAI's harness behind it. Standard · $9/mo Host opencode The open-source choice when you want to read the harness you are trusting with a shell. opencode is SST's agent, it is model-agnostic, and nothing about it is a black box. Standard · $9/mo Host Qwen Code The one to host if you are running Qwen or any OpenAI-compatible endpoint — including a self-hosted or regional provider that the other harnesses will not talk to. Standard · $9/mo Host Aider The only runtime here whose unit of work is a git commit. Aider edits a repo and commits, so the agent's output is a diff you can review and revert rather than a chat log you have to act on yourself. Standard · $9/mo Host goose Choose goose when the job needs tools that are not files and a shell. Its extension system is the point — it is the most natural fit for wiring an agent into other systems. Standard · $9/mo Host pi The lean option. pi does the coding-agent loop without a heavy harness around it, which makes it noticeably faster to first token on short tasks — the ones you actually send from a phone. Standard · $9/mo Host deepagents The one that plans before it acts. deepagents runs LangChain's deep-agent loop — an explicit plan, subagents, skills — which is what you want for a task measured in hours rather than a question measured in seconds. Standard · $9/mo Host Cursor Agent Host this if your team already lives in Cursor and you want the same agent reachable when nobody has the editor open.

Or just describe what you want

The builder's solutions engineer reads the same capability matrix these guides do, asks a few questions, and recommends a runtime. First agent month is free.

Describe your agent → See example agents
Showcase · Pricing · Developer API · ClawMetry blog