Pinky

// docs

## install

Linux / macOS

curl -fsSL https://moltctrl.com/install.sh | sh

Windows

irm https://moltctrl.com/install.ps1 | iex

Build from source

cargo install --git https://github.com/takschdube/moltctrl

## quick start

Run moltctrl with no arguments to launch the interactive wizard. Pinky will walk you through provider selection, authentication, and instance naming.

moltctrl
Select provider: Anthropic
Auth mode: api-key
API key: sk-ant-***
Instance name: my-agent
✓ Instance 'my-agent' created and started

## authentication

ModeDescriptionFlag
api-keyStandard API key authentication--api-key
max-tokenClaude Max/Pro subscription via setup token--setup-token

## commands

CommandDescription
moltctrlInteractive wizard
moltctrl createCreate a new instance
moltctrl start <name>Start an instance
moltctrl stop <name>Stop an instance
moltctrl restart <name>Restart an instance
moltctrl destroy <name>Remove an instance
moltctrl listList all instances
moltctrl status <name>Show instance status
moltctrl logs <name>View instance logs
moltctrl open <name>Open instance in browser
moltctrl update <name>Update instance config
moltctrl doctorDiagnose environment issues
moltctrl preconfig listList saved preconfigs
moltctrl preconfig show <n>Show preconfig details

## providers

Anthropic
ANTHROPIC_API_KEY
OpenAI
OPENAI_API_KEY
Google
GOOGLE_API_KEY
OpenRouter
OPENROUTER_API_KEY
AWS Bedrock
AWS_ACCESS_KEY_ID
Ollama
OLLAMA_HOST

Provider resolution order: CLI flags > preconfig > env vars > interactive prompt

## docker mode

Run agents in Docker containers with 15 security hardening measures including read-only root filesystem, no-new-privileges, dropped capabilities, and seccomp profiles.

moltctrl create --provider anthropic --docker --name secure-agent
✓ Docker instance 'secure-agent' created
moltctrl start secure-agent
✓ Container started with security hardening

## resource limits

FlagDefaultDescription
--mem512MMemory limit
--cpus1.0CPU cores
--pids256Max processes
--portautoGateway port