agentleFS
silencelensilencelen/huginn/skills/huginn/SKILL.md

huginn

Skill1 starsChanged 29 days agoMITView on GitHub

Huginn keeps Claude Code sessions running on a server the person owns; this machine reaches them with the `huginn` command, huginn core and the console, or the desktop and Android apps. Use when the person mentions huginn, huginn core, the huginn CLI or console, "my sessions on the server", devices, rounds, routes, the local AI tier, or asks what a `huginn ...` command does or where huginn keeps its files on this machine.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    The token is root on the server: Never print it, never put it in argv, never copy it out of ~/.config/huginn/secrets or …

    line 65
  2. Very specific:

    Something is broken?: Use the huginn-doctor skill: step 1 is huginn doctor --json.

    line 70

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Huginn on this machine

Huginn has two parts:

  • Huginn Server: one always-on Linux machine the person controls. Claude Code runs there in persistent tmux sessions, and huginn-appd (the daemon, port 8787, HTTPS) serves the clients.
  • Huginn Client: what is on THIS machine. The huginn command (a shell function from ~/.huginn/huginn.sh, or huginn.ps1 in PowerShell), huginn core (one Go binary, also called huginn, that runs as a per-user service, keeps the pairing with the server, draws the tray and the console, and can run work here as a device), and optionally the desktop app. The Android app connects to the same server.

Two programs answer to huginn: the shell function owns attaching and the session verbs (over SSH, Host huginn); the binary owns core, device, console, setup, doctor, skills and uninstall. Each hands the other what it does not own, so any spelling works.

Where things live here

WhatLinuxWindows
huginn core binary~/.local/bin/huginn -> ~/.local/share/huginn/current/huginn (versions under ~/.local/share/huginn/versions/)%LOCALAPPDATA%\huginn\bin\huginn.exe
the shell client~/.huginn/huginn.sh, loaded by ~/.bashrc (and ~/.zshrc, ~/.profile)~\.huginn\huginn.ps1, loaded by $PROFILE
core.json (pairing, routes, pins, device settings; never the token)~/.config/huginn/core.json%APPDATA%\huginn\core.json
the access token~/.config/huginn/secrets/ (0600 files; HUGINN_SECRETS=keyring uses the Secret Service)%APPDATA%\huginn\secrets\
state (markers, no data)~/.local/state/huginn/%LOCALAPPDATA%\huginn\state\
the core's socket$XDG_RUNTIME_DIR/huginn/core.sock%LOCALAPPDATA%\huginn\run\core.sock
the boot service~/.config/systemd/user/huginn-core.service, with lingerthe "Huginn Core" scheduled task; "Huginn Tray" task + Startup Huginn.lnk
logsjournalctl --user -u huginn-core.servicehuginn core status; Task Scheduler's history
these skills~/.local/share/huginn/skills/<name>, linked from ~/.claude/skills/<name>%LOCALAPPDATA%\huginn\skills\<name>, copied to ~\.claude\skills\<name>
local AI tier (optional)~/.config/huginn-local%ProgramData%\huginn-local
SSH to the serverHost huginn in ~/.ssh/config, key ~/.ssh/id_ed25519the same, under the profile

Nothing here is shared between accounts: every path is per user. Details: core.md.

The verbs, in one breath

  • Sessions (over SSH): huginn main, huginn <name>, huginn attach <name>, huginn ls, huginn status, huginn end <name>, huginn kill <name>, huginn archive, huginn revive.
  • The console: plain huginn in a terminal (huginn console). console.md.
  • This machine: huginn core status|install|start|pair|route|update, huginn device on|off|status, huginn setup ssh, huginn doctor [--json], huginn skills install|remove|status, huginn update, huginn uninstall.
  • The server, rendered there (over SSH, the token never leaves the server): huginn devices, huginn rounds, huginn headroom, huginn projects, huginn llm, huginn usage, huginn -p "question" (read-only), huginn -y "task" (with tools).
  • huginn help, and huginn <verb> --help for one verb, answered locally.

The full table with flags: verbs.md. Devices and scopes: devices.md. Local AI: local-ai.md.

How to behave

  • Read first. huginn core status, huginn device status, huginn skills status, huginn core route and huginn doctor --json change nothing. Prefer them to guessing.
  • Ask before anything that ends or changes work for the person: huginn end, huginn kill, huginn archive, huginn revive, huginn -y, huginn device on --scope work|own, huginn core pair, huginn core route remove|pin, huginn update, huginn uninstall, huginn local on|off. A bare huginn <name> attaches a terminal: it is for the person, not for you.
  • The token is root on the server. Never print it, never put it in argv, never copy it out of ~/.config/huginn/secrets or /etc/huginn-appd/token. huginn core pair reads it from stdin or a prompt for that reason.
  • Never edit core.json by hand while the core runs; use huginn core route and huginn device on (they go through the core's socket).
  • Something is broken? Use the huginn-doctor skill: step 1 is huginn doctor --json.
  • On the server itself, Claude Code sessions have the huginn-server skill instead; this one is about the client machine.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.