$ dpalv1.4.2Daemon Pal
A tiny process supervisor for macOS and Linux. Keep your servers, workers, and scripts running, and restart them when they exit. No root, no dependencies, one binary.
# Install with npm (or Homebrew, .deb, .rpm).
$ npm install -g @partoska/dpal# Start the control process.
$ dpal -C &# Add a process to supervise.
$ dpal -c /usr/bin/python3 -m http.server 8080# Start it and check its state.
$ dpal -s python3$ dpal -l -aWhy Daemon Pal?
Small and Dependency-Free
One C99 binary that needs nothing but the standard C library. It fits just as well on a Raspberry Pi Zero as on a build server.
Runs in User Space
No root and no system service files. Daemon Pal works next to systemd or launchd without touching them.
Restarts on Exit
When a supervised process exits, Daemon Pal starts it again. Set a restart delay per process so a failing service does not loop.
Follow the Logs
The stdout and stderr of every process go to rotating log files. Run dpal -L to print the last lines of one or all of them and follow new output as it arrives.
Plain INI Configuration
Define processes with arguments, environment variables, restart delays, and log rotation in a readable INI file. Import, export, and persist the list at any time.
Isolated Working Directories
Each control process owns its own working directory. Run one per project and control each one separately with the -D parameter.
Script-Friendly Output
List processes as JSON, CSV, or plain names with the -F parameter. Shell scripts, CI jobs, and AI agents can read the state without parsing a table.
Ready for AI Agents
Daemon Pal ships with an agent skill. Install it with one command, and AI coding agents such as Claude Code and Codex can manage your processes for you.
$ npx skills add partoska/dpalOne Binary, Two Modes
The same dpal binary runs as a long-lived control process or as a short-lived command executor. You start the control process once, then send it commands.
Control Process
The daemon that owns your processes. It starts them, watches them, restarts them when they exit, and writes their logs.
$ dpal -C -a &Command Executor
A short-lived client that sends one command to the running control process: list, start, stop, restart, add, delete, import, export, persist, or shut down. It also follows the process logs.
$ dpal -r webQuick Start
Go from install to a supervised process in five steps. The example serves the current directory over HTTP with Python.
Install dpal
Install the CLI and check the version. Prefer Homebrew or a native package? See the install options.
$ npm install -g @partoska/dpal$ dpal -vStart the control process
Run the daemon in the background. It stores its state in the directory set by the variable
$DPAL_HOME, or in~/.dpalwhen the variable is not set.$ dpal -C &Add a process
Give the executable path and its arguments. The process is named after the executable, here
python3.$ dpal -c /usr/bin/python3 -m http.server 8080Start it and check
Start the process by name or ID, then list all processes with their IDs and states.
$ dpal -s python3$ dpal -l -aKeep or shut down
Persist the list so it loads again on the next start, or stop the control process and every process it supervises.
$ dpal -p$ dpal -k
Examples
Describe your processes in an INI file, then load it and start everything in one command. Put comments on their own lines.
; Each [section] is one process, named after the section.[web]Id=1App=/usr/bin/nodeDir=/srv/web; First Arg is argv[0], then one Arg per argument.Arg=nodeArg=server.jsEnv=(Name=PORT,Value=8080)Env=(Name=NODE_ENV,Value=production); Wait 5 seconds before restarting after an exit.RestartSec=5; Rotate the log at 10 MB and keep 3 old files.LogMaxSize=10485760LogRotation=3 [worker]Id=2App=/usr/bin/python3Dir=/srv/workerArg=python3Arg=worker.pyRestartSec=10$ dpal -C -i ecosystem.ini -a &$ dpal -l -aMore options are in the documentation.
Install
Pick the method that fits your system. Every option installs the same dpal binary.
$ brew install partoska/tap/dpalFor macOS and Linux on x64 and ARM64.
Commands use the latest version, 1.4.2. See the release page for all files and SHA256 checksums.
Add the AI Agent Skill
Let AI coding agents such as Claude Code and Codex run dpal for you. The installer adds the Daemon Pal skill to the skills directory of your agent. Install dpal itself first.
$ npx skills add partoska/dpalCommand Reference
Run one command per call. You can target processes by name or ID, or all of them with the -a parameter. Run dpal -h for the built-in help.
Control process
- -C, --control
- Start the control process.
- -D, --dir
- Use this working directory.
- -i, --ini
- Load processes from an INI file.
- -a, --auto-start
- Start every loaded process.
- -f, --force
- Start or run even if the working directory is locked.
Command executor
- -l, --list
- List processes.
- -s, --start
- Start processes.
- -t, --stop
- Stop processes with
SIGTERM. - -r, --restart
- Stop and start processes again.
- -c, --create
- Add a process from an executable and its arguments.
- -d, --delete
- Remove processes from the list.
- -i, --import
- Import processes from an INI file.
- -e, --export
- Export processes to an INI file.
- -p, --persist
- Save the list so it loads on the next start.
- -k, --kill
- Stop the control process and all its processes.
- -L, --log
- Print the last log lines and follow new output.
- -a, --all
- Apply the command to all processes.
- -n, --lines
- Number of last lines per log that
-Lprints (default 10). - --no-follow
- Make
-Lprint the last lines and exit. - -F, --format
- Output format for
-l:plain,json,csv, orone.
Use Cases
Servers and Devices
- →Keep services running on a Raspberry Pi or other small device.
- →Supervise web apps and workers on a server without root access.
- →A lightweight alternative to
systemdunits for user-owned processes.
Development and Automation
- →Start your whole local stack (API, workers, watchers) with one command.
- →Separate process groups per project with their own working directories.
- →Let an AI coding assistant start, stop, and restart your dev servers.
Technical Specifications
Keep your processes running.
Install Daemon Pal in one command and supervise your first process in under a minute.
View the source on GitHub, get the package on npm, read the documentation, or add the AI agent skill.