Skip to main content

Python SDK for CMDOP agent interaction

Project description

cmdop

Turn any machine into an API endpoint.

PyPI Python License

from cmdop import CMDOPClient

# Server in another country. Behind NAT. Behind firewall. Don't care.
with CMDOPClient.remote(api_key="cmd_xxx") as server:
    server.terminal.execute("docker restart app")
    server.files.write("/etc/nginx/sites/new.conf", config)
    logs = server.files.read("/var/log/app/error.log")

No SSH. No VPN. No open ports. No bullshit.


The Problem

You need to run a command on a remote server. What do you do?

Option A: SSH

→ Generate keys
→ Copy public key to server
→ Configure sshd
→ Open port 22
→ Hope firewall allows it
→ Pray NAT doesn't break it
→ paramiko/fabric in Python
→ Debug why it doesn't work

Option B: Build an API

→ Write FastAPI endpoints
→ Implement authentication
→ Get SSL certificate
→ Open ports
→ Deploy to every server
→ Maintain all of this
→ Write client code
→ Repeat for each new operation

Option C: CMDOP

server.terminal.execute("your command")

That's it.


How It Works

┌─────────────┐      ┌─────────────┐      ┌─────────────┐
│  Your Code  │ ──── │ Cloud Relay │ ──── │   Agent     │
│   (SDK)     │      │grpc.cmdop.com│     │  (on server)│
└─────────────┘      └─────────────┘      └─────────────┘
                            │
        Outbound only ──────┘
        No open ports
        Works through any NAT/firewall
  1. Install agent on target machine
  2. Agent connects outbound to cloud relay
  3. Your code connects to same relay with API key
  4. Commands flow through, results come back
  5. Zero network configuration required

What Can You Do?

Terminal — full shell access:

session = server.terminal.create(shell="/bin/bash")
server.terminal.send_input(session.session_id, "kubectl get pods\n")
output = server.terminal.get_history(session.session_id)

Files — complete filesystem control:

server.files.list("/home/user")
server.files.read("/etc/nginx/nginx.conf")
server.files.write("/tmp/config.json", b'{"key": "value"}')
server.files.delete("/tmp/garbage", recursive=True)
server.files.copy("/backup/db.sql", "/restore/db.sql")

AI Agents — run LLM agents with type-safe responses:

from pydantic import BaseModel

class ServerHealth(BaseModel):
    status: str
    cpu_percent: float
    memory_used_gb: float
    disk_free_gb: float
    issues: list[str]

# Run agent, get typed response
result = server.agent.run(
    prompt="Check server health and report issues",
    output_schema=ServerHealth
)
health: ServerHealth = result.output  # Fully typed!
if health.issues:
    alert(health.issues)

Real World

Give AI Agents Hands

from pydantic import BaseModel

class DeployResult(BaseModel):
    success: bool
    version: str
    containers_running: int
    errors: list[str]

# Agent executes commands, parses output, returns structured data
with CMDOPClient.remote(api_key=KEY) as server:
    result = server.agent.run(
        prompt="Deploy myapp:v2.1 and verify all containers are healthy",
        agent_type="terminal",
        output_schema=DeployResult
    )
    if not result.output.success:
        rollback(result.output.errors)

Deploy Without The Circus

# No Ansible. No Terraform. No YAML hell.
def deploy(version: str):
    with CMDOPClient.remote(api_key=PROD_KEY) as prod:
        prod.terminal.execute(f"docker pull myapp:{version}")
        prod.terminal.execute("docker-compose up -d")

        # Verify
        health = prod.files.read("/var/log/myapp/health.log")
        assert b"healthy" in health

Fleet Management

# Update 1000 edge devices
async def update_fleet(device_keys: list[str], new_config: bytes):
    async with asyncio.TaskGroup() as tg:
        for key in device_keys:
            tg.create_task(update_device(key, new_config))

async def update_device(key: str, config: bytes):
    async with AsyncCMDOPClient.remote(api_key=key) as device:
        await device.files.write("/etc/app/config.yml", config)
        await device.terminal.execute("systemctl restart app")

Debug Customer Issues

# Support engineer gets temporary access
def diagnose(customer_key: str):
    with CMDOPClient.remote(api_key=customer_key) as machine:
        # See what's running
        session = machine.terminal.create()
        machine.terminal.send_input(session.session_id, "ps aux\n")

        # Read their logs
        logs = machine.files.read("~/Library/Logs/MyApp/error.log")

        # Check disk space
        machine.terminal.send_input(session.session_id, "df -h\n")

Installation

pip install cmdop

Quick Start

from cmdop import CMDOPClient

# Remote server via cloud relay
with CMDOPClient.remote(api_key="cmd_xxx") as client:
    result = client.files.list("/home")
    print(result.entries)

# Local agent on same machine
with CMDOPClient.local() as client:
    result = client.files.list("/home")

Async

from cmdop import AsyncCMDOPClient

async with AsyncCMDOPClient.remote(api_key="cmd_xxx") as client:
    await client.terminal.execute("echo hello")
    content = await client.files.read("/etc/hostname")

API

Terminal

Method What it does
create(shell, cols, rows) Start terminal session
send_input(id, data) Send commands/keystrokes
get_history(id) Get output
resize(id, cols, rows) Resize terminal
send_signal(id, signal) Send SIGINT/SIGTERM/etc
close(id) End session

Files

Method What it does
list(path) List directory
read(path) Read file
write(path, content) Write file
delete(path) Delete file/dir
copy(src, dst) Copy
move(src, dst) Move/rename
mkdir(path) Create directory
info(path) Get metadata

Agent

Method What it does
run(prompt, output_schema=None) Run AI agent, optionally with Pydantic model for structured output

Supported agent types: chat, terminal, command, router, planner

Structured output with Pydantic:

class MyResult(BaseModel):
    answer: str
    confidence: float

result = client.agent.run("...", output_schema=MyResult)
typed_data: MyResult = result.output  # Validated against schema

Security

  • TLS everywhere — all traffic encrypted
  • Outbound only — agent never accepts incoming connections
  • API key scoping — keys bound to specific agents/teams
  • No credentials on wire — no SSH keys, no passwords
  • Audit logs — every action tracked

Requirements

  • Python 3.10+
  • CMDOP agent running on target machine

Links

cmdop.com

License

MIT

Project details


Download files

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

Source Distribution

cmdop-0.1.6.tar.gz (126.7 kB view details)

Uploaded Source

Built Distribution

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

cmdop-0.1.6-py3-none-any.whl (224.5 kB view details)

Uploaded Python 3

File details

Details for the file cmdop-0.1.6.tar.gz.

File metadata

  • Download URL: cmdop-0.1.6.tar.gz
  • Upload date:
  • Size: 126.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.10.18

File hashes

Hashes for cmdop-0.1.6.tar.gz
Algorithm Hash digest
SHA256 ddc2a1846c98997576a10d4ad4b35ce1dc856f1bf09aeeeac7579610d1efa9c9
MD5 0407c9fde0db8204cf26047443828529
BLAKE2b-256 661ac4483bd003cc51d70733e14db42becabcc8ec145749f90c14a53fc257ceb

See more details on using hashes here.

File details

Details for the file cmdop-0.1.6-py3-none-any.whl.

File metadata

  • Download URL: cmdop-0.1.6-py3-none-any.whl
  • Upload date:
  • Size: 224.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.10.18

File hashes

Hashes for cmdop-0.1.6-py3-none-any.whl
Algorithm Hash digest
SHA256 41ab1e7a0729cb941a702d74a54cb6ec6987c3abfa46b1c359f2c81094f7fa60
MD5 888b5fcf426764dc27aebdfdf58fe4ef
BLAKE2b-256 25b05341252c8dc2adfe73e25fe082a623c868541a32228374131851ce9a1527

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 Pingdom Monitoring Sentry Error logging StatusPage Status page