agent

stakpak up        # alias for: stakpak autopilot up
stakpak down      # alias for: stakpak autopilot down

You can also use the canonical subcommands:

stakpak autopilot up
stakpak autopilot status
stakpak autopilot logs
stakpak autopilot down
stakpak autopilot doctor

Autopilot prerequisites

Before running autopilot on a remote VM:

• Docker must be installed and accessible to the current user
• 2GB+ RAM is recommended for reliable autopilot + sandbox runs
• Swap is strongly recommended on small Linux hosts
• Linux user services may require linger to survive logout

stakpak up now runs preflight checks before startup, and stakpak autopilot doctor can be used as a deployment-readiness check before first boot:

stakpak autopilot doctor
stakpak up

See also: cli/README.md

Unified configuration (profiles + autopilot wiring)

• ~/.stakpak/config.toml: profile behavior (model, allowed_tools, auto_approve, system_prompt, max_turns, provider credentials)
• ~/.stakpak/autopilot.toml: runtime wiring (schedules, channels, notification routes, service/server settings)

Use profile = "name" on schedules/channels and keep behavior inside profile definitions. Profiles control how the agent behaves: model, allowed tools, auto-approval, system prompt, max turns, and provider credentials. Notification routes only control where messages are delivered.

Schedule and channel profiles are intentionally separate:

• schedule --profile monitoring: behavior for runs started by that schedule
• channel --profile ops: behavior for sessions started from inbound Slack/Telegram/Discord messages
• notification --target "#ops": where schedule notifications are sent; it does not choose the model or tools

Notification routing uses two words everywhere:

• channel: the transport, such as slack, telegram, or discord
• target: the destination inside that transport, such as Slack #ops or C1234567890

autopilot channel add ... --target sets the default notification route. Schedules inherit it unless you add --notify-target or --notify-channel.

# default notification route: slack:#ops
stakpak autopilot channel add slack --bot-token "$SLACK_BOT_TOKEN" --app-token "$SLACK_APP_TOKEN" --profile ops --target "#ops"

# schedule runs with the monitoring profile and inherits the default route
stakpak autopilot schedule add health --cron '*/5 * * * *' --prompt 'Check health' --profile monitoring

# schedule still runs with monitoring, but notifies a different Slack target
stakpak autopilot schedule add deploy-watch --cron '*/15 * * * *' --prompt 'Watch deploys' --profile monitoring --notify-target "#deploys"

In ~/.stakpak/autopilot.toml, the same setup is represented as:

[notifications]
channel = "slack"
target = "#ops"

[[schedules]]
name = "health"
cron = "*/5 * * * *"
prompt = "Check health"
profile = "monitoring"

[[schedules]]
name = "deploy-watch"
cron = "*/15 * * * *"
prompt = "Watch deploys"
profile = "monitoring"
notify_target = "#deploys"

Slack public channel names such as #ops are accepted where Slack supports them. Channel IDs are most reliable for private channels, DMs, and scripts.

Full setup guide: cli/README.md

🔒 Security Hardened

• Mutual TLS (mTLS) - End-to-end encrypted MCP
• Dynamic Secret Substitution - AI can read/write/compare secrets without seeing actual values
• Secure Password Generation - Generate cryptographically secure passwords with configurable complexity
• Privacy Mode - Redacts sensitive data like IP addresses and AWS account IDs

🛠️ Built for DevOps Work

• Asynchronous Task Management - Run background commands like port forwarding and servers with proper tracking and cancellation
• Real-time Progress Streaming - Long-running processes (Docker builds, deployments) stream progress updates in real-time
• Infrastructure Code Indexing - Automatic local indexing and semantic search for Terraform, Kubernetes, Dockerfile, and GitHub Actions
• Documentation Research Agent - Built-in web search for technical documentation, cloud providers, and development frameworks
• Subagents - Specialized research agents for code exploration and sandboxed analysis with different tool access levels (enabled with --enable-subagents flag)
• Bulk Message Approval - Approve multiple tool calls at once for efficient workflow execution
• Reversible File Operations - All file modifications are automatically backed up with recovery capabilities

🧠 Adaptive Intelligence

• Rule Books - Customize agent behavior with internal standard operating procedures, playbooks, and organizational policies
• Persistent Knowledge - Agent learns from interactions, remembers incidents, resources, and environment details to adapt to your workflow

Installation

All installation options (Linux, MacOs, Windows)

Check the docs

Homebrew (Linux & MacOS)

brew tap stakpak/stakpak
brew install stakpak

To update it you can use

brew update
brew upgrade stakpak

Binary Release

Download the latest binary for your platform from our GitHub Releases.

Docker

This image includes the most popular CLI tools the agent might need for everyday DevOps tasks like docker, kubectl, aws cli, gcloud, azure cli, and more.

docker pull ghcr.io/stakpak/agent:latest

Usage

You can use your own Anthropic or OpenAI API keys, custom OpenAI compatible endpoint, or a Stakpak API key.

Option A: Running with a Stakpak API Key (no card required)

Just run stakpak and follow the instructions which will create a new API key for you.

stakpak

Brave users may encounter issues with automatic redirects to localhost ports during the API key creation flow. If this happens to you:

Copy your new key from the browser paste it in your terminal

Non-interactive setup (CI/scripts)

stakpak auth login --api-key $STAKPAK_API_KEY

Or set the environment variable

export STAKPAK_API_KEY=<mykey>

View current account (Optional)

stakpak account

Option B: Running Without a Stakpak API Key

Non-interactive setup (CI/scripts)

# Anthropic
stakpak auth login --provider anthropic --api-key $ANTHROPIC_API_KEY

# OpenAI
stakpak auth login --provider openai --api-key $OPENAI_API_KEY

# Gemini
stakpak auth login --provider gemini --api-key $GEMINI_API_KEY

Manual configuration

Create ~/.stakpak/config.toml with one of these configurations:

Option 1: Bring Your Own Keys (BYOK) - Use your Anthropic/OpenAI API keys:

[profiles.byok]
provider = "local"

# Unified model preference field
model = "anthropic/claude-sonnet-4-5"

# Built-in providers - credentials can also be set via environment variables
# (ANTHROPIC_API_KEY, OPENAI_API_KEY, GEMINI_API_KEY)
[profiles.byok.providers.anthropic]
type = "anthropic"
api_key = "sk-ant-..."

[profiles.byok.providers.openai]
type = "openai"
api_key = "sk-..."

[profiles.byok.providers.gemini]
type = "gemini"
api_key = "..."

[settings]

Option 2: Bring Your Own LLM - Use a local OpenAI-compatible endpoint (e.g. Ollama, LM Studio):

[profiles.offline]
provider = "local"

# Custom provider models use the format: provider_key/model_name
model = "offline/qwen/qwen3-coder-30b"

# The provider key "offline" becomes the model prefix
[profiles.offline.providers.offline]
type = "custom"
api_endpoint = "http://localhost:11434/v1"
# api_key is optional for local providers

[settings]

Option 3: Mix Built-in and Custom Providers:

[profiles.hybrid]
provider = "local"

# Unified model field (provider-prefixed)
model = "anthropic/claude-sonnet-4-5"

[profiles.hybrid.providers.anthropic]
type = "anthropic"
# Uses ANTHROPIC_API_KEY env var

[profiles.hybrid.providers.offline]
type = "custom"
api_endpoint = "http://localhost:11434/v1"

[settings]

Stakpak applies available updates in the background during interactive startup. When an update is applied, restart any long-running Stakpak processes to pick up the new binary.

Then run with your profile:

stakpak --profile byok
# or
stakpak --profile offline
# or
stakpak --profile hybrid

Start Stakpak Agent TUI

# Open the TUI
stakpak
# Resume execution from a checkpoint
stakpak -c <checkpoint-id>

Start Stakpak Agent TUI with Docker

docker run -it --entrypoint stakpak ghcr.io/stakpak/agent:latest
# for containerization tasks (you need to mount the Docker socket)
docker run -it \
   -v "/var/run/docker.sock":"/var/run/docker.sock" \
   -v "{your app path}":"/agent/" \
   --entrypoint stakpak ghcr.io/stakpak/agent:latest

MCP Modes

You can use Stakpak as a secure MCP proxy or expose its security-hardened tools through an MCP server.

MCT Server Tools

• Local Mode (--tool-mode local) - File operations and command execution only (no API key required)
• Remote Mode (--tool-mode remote) - AI-powered code generation and search tools (API key required)
• Combined Mode (--tool-mode combined) - Both local and remote tools (default, API key required)

Start MCP Server

# Local tools only (no API key required, mTLS enabled by default)
stakpak mcp start --tool-mode local

# Remote tools only (AI tools optimized for DevOps)
stakpak mcp start --tool-mode remote

# Combined mode (default - all tools with full security)
stakpak mcp start

# Disable mTLS (NOT recommended for production)
stakpak mcp start --disable-mcp-mtls

Additional flags for the MCP server:

• --disable-secret-redaction – not recommended; prints secrets in plaintext to the console
• --privacy-mode – redacts additional private data like IP addresses and AWS account IDs
• --enable-slack-tools – enables experimental Slack tools

MCP Proxy Server

Stakpak also includes an MCP proxy server that can multiplex connections to multiple upstream MCP servers using a configuration file.

# Start MCP proxy with automatic config discovery
stakpak mcp proxy

# Start MCP proxy with explicit config file
stakpak mcp proxy --config-file ~/.stakpak/mcp.toml

# Disable secret redaction (NOT recommended – secrets will be printed in logs)
stakpak mcp proxy --disable-secret-redaction

# Enable privacy mode to redact IPs, account IDs, etc.
stakpak mcp proxy --privacy-mode

Configure External MCP Servers

External MCP servers are configured in mcp.toml (or mcp.json) and loaded by stakpak mcp proxy.

Config file discovery order:

1. ~/.stakpak/mcp.toml or ~/.stakpak/mcp.json
2. .stakpak/mcp.toml or .stakpak/mcp.json (from current directory)
3. ./mcp.toml or ./mcp.json

Add servers via CLI:

# Add a stdio MCP server
stakpak mcp add context7 --command npx --args "-y,@upstash/context7-mcp"

# Add a stdio MCP server with environment variables (repeat --env)
stakpak mcp add internal-tools --command npx --args "-y,@acme/internal-mcp" --env "MCP_API_KEY=<token>" --env "MCP_REGION=us-east-1"

# Add an HTTP MCP server
stakpak mcp add github --url https://api.githubcopilot.com/mcp --headers "Authorization=Bearer <token>"

# Inspect and manage servers
stakpak mcp list
stakpak mcp get github
stakpak mcp disable github
stakpak mcp enable github

# Start the proxy with configured external MCPs
stakpak mcp proxy

Or define servers directly in ~/.stakpak/mcp.toml:

[mcpServers.context7]
command = "npx"
args = ["-y", "@upstash/context7-mcp"]
env = { MCP_API_KEY = "<token>", MCP_REGION = "us-east-1" }

[mcpServers.github]
url = "https://api.githubcopilot.com/mcp"
headers = { Authorization = "Bearer <token>" }

[mcpServers.optional]
command = "npx"
args = ["-y", "some-mcp-server"]
disabled = true

Use disabled = true to keep a server configured without loading it.

Agent Client Protocol (ACP)

ACP is a standardized protocol that enables AI agents to integrate directly with code editors like Zed, providing seamless AI-powered development assistance.

What ACP Offers with Stakpak

• Real-time AI Chat - Natural language conversations with context-aware AI assistance
• Live Code Analysis - AI can read, understand, and modify your codebase in real-time
• Tool Execution - AI can run commands, edit files, search code, and perform development tasks
• Session Persistence - Maintains conversation context across editor sessions
• Streaming Responses - Real-time AI responses with live progress updates
• Agent Plans - Visual task breakdown and progress tracking

Installation & Setup

1. Install Stakpak (if not already installed)
2. Configure Zed Editor - Add to ~/.config/zed/settings.json:

{
  "agent_servers": {
    "Stakpak": {
      "command": "stakpak",
      "args": ["acp"],
      "env": {}
    }
  }
}

1. Start ACP Agent:

stakpak acp

1. Use in Zed - Click Assistant (✨) → + → New stakpak thread

Rulebook Management

Manage your standard operating procedures (SOPs), playbooks, and runbooks with Stakpak Rulebooks. Rulebooks customize agent behavior and provide context-specific guidance.

# List all rulebooks
stakpak rulebooks get
# or use the short alias
stakpak rb get

# Get a specific rulebook
stakpak rb get stakpak://my-org/deployment-guide.md

# Create or update a rulebook from a markdown file
stakpak rb apply my-rulebook.md

# Delete a rulebook
stakpak rb delete stakpak://my-org/old-guide.md

Rulebook Format

Rulebooks are markdown files with YAML frontmatter:

---
uri: stakpak://my-org/deployment-guide.md
description: Standard deployment procedures for production
tags:
  - deployment
  - production
  - sop
---

# Deployment Guide

Your deployment procedures and guidelines here...

Shell Completion

Enable tab-completion for every stakpak subcommand, flag, and argument.

# Bash
echo 'source <(stakpak completion bash)' >> ~/.bashrc

# Elvish
echo 'eval (stakpak completion elvish | slurp)' >> ~/.elvish/rc.elv

# Fish
echo 'stakpak completion fish | source' > ~/.config/fish/completions/stakpak.fish

# Zsh
echo 'source <(stakpak completion zsh)' >> ~/.zshrc

# PowerShell
Add-Content -Path $PROFILE -Value 'stakpak completion powershell | Out-String | Invoke-Expression'

Supported shells: bash, elvish, fish, powershell, zsh.

Shell Mode

Execute system commands explicitly from the input bar.

Check the Shell Mode docs for details on background vs foreground execution.

Platform Testing

Windows

Comprehensive testing report for Windows CLI functionality, including installation, configuration, and integration with WSL2 and Docker.

View Windows Testing Report

⭐ Like what we're building?

If our Agent saves you time or makes your DevOps life easier,
consider giving us a star on GitHub — it really helps!