agentleFS
Noah-OSNoah-OS/longhaul/skills/manage/SKILL.md

manage

Skill4 starsChanged 24 days agoMITView on GitHub

Run Longhaul's setup from this chat: accounts (add, sign in, remove, pick a chat's), switching, subagent models, the VS Code panel, the status line, auto-update, clearing limits, uninstalling. Each change runs only after the user's yes.

8 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions8 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 8

    1. vercel/eve★ 5.9ksame name
    2. nicholasglazer/gnosis-mcp★ 30same name
    3. kinlace/parent-recap★ 8same name
    4. monoscope-tech/skills★ 4same name
    5. divanc/skills★ 3same name
    6. jskswamy/skills★ 3same name
    7. getlark/skills★ 1same name
    8. AvyanshKatiyar/openai-ads-mcp★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository3 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    Run each by that full path, ${CLAUDE_PLUGIN_ROOT}/bin/longhaul.

    line 9
  2. Distinctive:

    Ask first, always. For every command that changes something, show the user the exact command, say in plain words what it changes, ask, and …

    line 13
  3. Very specific:

    Add one: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul add NAME EMAIL, with --repo PATH for each repository it alone may be used in (and those inside …

    line 26
  4. Very specific:

    Remove what Longhaul set up outside the plugin (its status line, VS Code's line, auto-update, the terminal command): …

    line 58
  5. Distinctive:

    If the shim refuses a command, show its message and do not try another form of it.

    line 72

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “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

Longhaul's setup actions, which once needed a terminal, run from this chat on the user's word: each line below is one action and its exact command. Run each by that full path, ${CLAUDE_PLUGIN_ROOT}/bin/longhaul; the plugin's shim runs these forms and refuses any other. What the user asked for: $ARGUMENTS (or the conversation before).

Ask first, always. For every command that changes something, show the user the exact command, say in plain words what it changes, ask, and run it only after their yes in this chat: one yes, one command. Claude Code's own permission prompt is not enough: the VS Code panel can run in a mode that approves Bash by itself. The lines marked "reads" change nothing and need no yes. Show what a command prints as it is.

Before ${CLAUDE_PLUGIN_ROOT}/bin/longhaul uninstall --yes --purge, say first that it deletes ~/.longhaul: the checkpoints, the usage ledger and the accounts list, for good.

Accounts:

  • See them, their sign-ins and limits, and the switching mode (reads): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul status
  • Add one: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul add NAME EMAIL, with --repo PATH for each repository it alone may be used in (and those inside it) and --dir DIR for its folder (default ~/.claude-<name>). It writes Longhaul's accounts list, then signs the account in.
  • Sign one in again: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul setup NAME --relogin; ${CLAUDE_PLUGIN_ROOT}/bin/longhaul setup NAME signs in one that is not.
  • Take one off the list: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul remove NAME; its folder and sign-in stay. Deleting its folder as well is not run from here: tell the user first that its sign-in and the chat history kept in that folder go for good, then that they do it themselves, with Remove and delete its sign-in folder… in /accounts or longhaul remove NAME --delete-folder in a terminal, which says what goes and asks first.
  • Where one may be used (reads): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul folders NAME. Change it: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul folders NAME --add PATH (a folder it may be used in, and those inside it), --remove PATH, each PATH starting with / or ~/ and repeatable, or ${CLAUDE_PLUGIN_ROOT}/bin/longhaul folders NAME --anywhere (drops its list). A folder on an account's list is kept for the accounts that list it, so say which other accounts that leaves out, as the command's output does.

Which account a chat runs on, and which a folder prefers:

  • Run this chat on one account from now on: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul use NAME. A VS Code panel chat moves the next time Claude is idle in it; any other chat starts there when it is next opened, and the command says which. It refuses an account not allowed in this chat's folder, one at its usage limit (${CLAUDE_PLUGIN_ROOT}/bin/longhaul clear NAME forgets a limit that is over), or one not signed in as expected. It is no lock: at a usage limit the switching mode decides, as for any chat.
  • Prefer an account in a folder: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul folders NAME --prefer PATH; ${CLAUDE_PLUGIN_ROOT}/bin/longhaul folders NAME --no-prefer PATH takes it off. A new chat in PATH or a folder inside it, and a chat there that moves at a usage limit, goes to NAME first while it has room; a reopened chat goes back to the account it ran on. A PATH the account may not be used in is refused, with why.

For add and setup, tell the user first that the sign-in opens in the browser and that they sign in there with that account's email: Longhaul accepts no other. A browser signed in to claude.ai as someone else: copy the link into a private window. Never ask for a password, a code or a token in the chat. If no browser opens, give the user the same line to type in a terminal, longhaul in place of the full path.

Settings:

  • Switching at a usage limit (reads): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul switch. Set it: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul switch MODE, MODE ask (ask first: in the terminal, and in the panel with Longhaul's pane, whose question only the user answers), auto (move by itself) or never (wait for the reset).
  • Subagent models (reads): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul models. Set them: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul models MODE (guide, on, off or trial), ${CLAUDE_PLUGIN_ROOT}/bin/longhaul models cap MODEL (haiku, sonnet, opus or fable; off for none), ${CLAUDE_PLUGIN_ROOT}/bin/longhaul models TYPE MODEL (one agent type's row).
  • The VS Code panel on Longhaul's accounts: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul vscode on --yes (it replaces any other program the panel was set to run); back to its own sign-in: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul vscode off. Each edits VS Code's settings, a copy kept. After either, tell the user to reload VS Code: Cmd+Shift+P, then Developer: Reload Window.
  • The status line: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul statusline install --yes sets Claude Code's to Longhaul's, replacing one of the user's own (say so; a copy is kept); ${CLAUDE_PLUGIN_ROOT}/bin/longhaul statusline uninstall puts the one before back.
  • Claude Code's auto-update for Longhaul (reads): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul update status. Set it: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul update on --yes or ${CLAUDE_PLUGIN_ROOT}/bin/longhaul update off --yes.
  • Forget the recorded usage limits (an account was reset before its time and Longhaul still waits for it): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul clear. The chats waiting for one carry on within a few seconds. One account's alone: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul clear NAME.

Something stuck (a wait or a question that never goes, a limit that is not a time, an accounts.json that will not read, the VS Code panel pointing at a Longhaul that is gone): offer this before any uninstall.

  • What is stuck, one line each with what its fix does (reads): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul repair --dry-run. Show it as it is.
  • The fixes do not run from here: tell the user to run longhaul repair in a terminal, which asks before each fix (longhaul repair --yes makes them all).

Uninstall:

  • What it would remove (reads): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul uninstall --dry-run; with ~/.longhaul too: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul uninstall --dry-run --purge. Show it before asking.
  • Remove what Longhaul set up outside the plugin (its status line, VS Code's line, auto-update, the terminal command): ${CLAUDE_PLUGIN_ROOT}/bin/longhaul uninstall --yes; ~/.longhaul too: ${CLAUDE_PLUGIN_ROOT}/bin/longhaul uninstall --yes --purge. The output says how to remove the plugin itself.

Terminal only: give the user the line to type there. The trials (longhaul trial ...: a chat must not know or set its own arm), longhaul advisor record-probe (half of the Advisor's gate, the owner's alone), and starting a chat (longhaul, -c, -r: in VS Code the panel already runs the chat on the right account).

With buttons: /accounts, typed by the user, does the account actions, switching, the VS Code panel and forgetting the limits in a pane, and is where the user answers the question at a usage limit. Tell the user of it when they would rather press than type; the rules above still hold for every command you run.

If the shim refuses a command, show its message and do not try another form of it.

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.