Skip to main content

nanoclaw

cmdop — messenger for AI agents

Installer for the cmdop CLI — one binary that runs an AI agent inside a container you own.

pip install nanoclaw
nanoclaw

That's it. nanoclaw detects your platform, downloads the cmdop binary, verifies it against the published checksums, installs it, and hands over.

Containers make agents reproducible, and they make them awkward: the agent has to be reachable from outside the container without opening the container to the world.

cmdop is a single static-ish executable with no runtime dependencies, which makes it a well-behaved thing to put in an image. This package installs it from within one, or from a plain shell.

How does CMDOP compare? CMDOP Skills vs OpenClaw: two approaches to AI agent skills →

Small enough to belong in an image

No dependency tree to resolve at build time and no service to link — the install is a download, a checksum verification, and a file in ~/.local/bin. That keeps layers small and rebuilds fast.

The same binary serves its own console, so the container needs no sidecar to be usable.

What is cmdop

One executable that is the agent, the relay, the local server and the web console at once. There is no gateway, backend, database server or broker to stand up — you add one binary to a machine you already own.

  • run an AI agent on your own machine, with your own tools
  • reach it from anywhere: terminal, files, remote execution
  • connect machines into a fleet
  • works with Claude Code and other agent engines under the hood

Full documentation: docs.cmdop.com

Usage

nanoclaw                    # install cmdop (if needed), then run it
nanoclaw --install-only     # install, don't hand over
nanoclaw --prefix ~/bin     # choose the install directory
nanoclaw --force            # reinstall even if already present
nanoclaw --quiet            # print only the installed path
nanoclaw -- --help          # everything after `--` goes to cmdop

Default install directory is ~/.local/bin (%LOCALAPPDATA%\cmdop\bin on Windows). No sudo, no system directories.

What it does, precisely

  1. Detects your OS and architecture.
  2. Fetches SHA256SUMS from install.cmdop.com.
  3. Downloads the matching binary to a temporary file.
  4. Verifies the SHA-256 against the published checksum. On mismatch it aborts and installs nothing.
  5. Moves the verified binary into the install directory and marks it executable.
  6. Runs it.

Supported: macOS (Apple Silicon, Intel), Linux (x64, arm64), Windows (x64).

What it does not do

  • No side effects at install time. pip install nanoclaw only unpacks files. Nothing is downloaded or executed until you run nanoclaw.
  • No network access on import. import nanoclaw touches nothing.
  • No telemetry. The package sends no analytics, ever.
  • No dependencies. Standard library only, so there is nothing to conflict with what you already have installed.

Python API

The installer is importable if you would rather script it:

import nanoclaw

target = nanoclaw.detect()
print(target.label, target.asset)

result = nanoclaw.install()
print(result.path, result.sha256)

nanoclaw.install() raises nanoclaw.InstallError with a user-facing message on any failure, and nanoclaw.UnsupportedPlatform when there is no published binary for the current machine.

Links

License

MIT

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

nanoclaw-2.0.1.tar.gz (9.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

nanoclaw-2.0.1-py3-none-any.whl (10.0 kB view details)

Uploaded Python 3

File details

Details for the file nanoclaw-2.0.1.tar.gz.

File metadata

  • Download URL: nanoclaw-2.0.1.tar.gz
  • Upload date:
  • Size: 9.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.2

File hashes

Hashes for nanoclaw-2.0.1.tar.gz
Algorithm Hash digest
SHA256 7d55a48032ea5e8c693b05b9b4517db3798edc6284d58ff6fbf6c15ee15f8818
MD5 8f825c575e96c767c072e52ec7b099f5
BLAKE2b-256 c0911302017f0c3084411f9da7b2a112181a97c9f8f3ccb848a63395c22912a8

See more details on using hashes here.

File details

Details for the file nanoclaw-2.0.1-py3-none-any.whl.

File metadata

  • Download URL: nanoclaw-2.0.1-py3-none-any.whl
  • Upload date:
  • Size: 10.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.2

File hashes

Hashes for nanoclaw-2.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 e366577b52cf13eb02aac1b4d1bf98e5ea6c1539540c55c0dbf1e1c1557a472a
MD5 4c73a5c54fe759a28c561db066e05b95
BLAKE2b-256 e619d9c007fb0afd47969a69e6ed937910923b90b3ae82856e241c169c7f2684

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page