Files
marketplace/cli/internal/scaffold/assets/commands/kettle/init.md
T
naudachu 8b1b11001a feat: drop the kettle plugin; the binary writes its own skills
The plugin and the binary shipped on two release cadences and nothing on an
operator's machine ever checked that the one they installed described the other.
The generated flag block existed precisely so a renamed flag could not ship with
documentation recommending the old one — and then shipped one version behind the
registry it came from, which is the same bug one hop downstream.

So the prose moved into the binary. `internal/scaffold` embeds every document;
`kettle init` and `kettle gen scaffold` write them into a project's own
`.claude/`. The two cannot disagree because there is one artefact.

The namespace survived the move. A project's skills are flat, so the prefix is
spelled into the directory name (`kettle-issue`); a project's *commands* take
their namespace from a subdirectory, so `commands/kettle/init.md` is still
`/kettle:init`. Four of the six command files are thin pointers at a skill, and
that is what kept ~1,600 lines of `/kettle:…` cross-references true without a
rewrite. `init` and `auth` lost `disable-model-invocation: true` — being a
command is that property — and `auth` now restricts `allowed-tools` so a model
cannot reach `kettle auth add` at all.

`gen scaffold` writes files whole rather than splicing a region. The old
refusal protected somebody's hand-written prose around the block; that prose is
embedded now, so there is none to protect, and preserving local edits would
freeze a project's documentation at whatever version first initialized it.
`--check` warns before an upgrade discards one.

The plugin's `agents-sync.sh` — 141 lines of Python behind a filename that said
`.sh` — became `internal/mirror` and `kettle mirror`. Same seven branches, same
refusal to merge two real files that differ, now with a table test per branch
and a check that a repair converges in one pass. `--hook` is the PreToolUse
form and exits 0 on every path including a panic. It is opt-in per project,
which is strictly narrower than the plugin hook that was on for everybody who
installed it.

`kettle init --interactive` walks a person through the login, the token (read
with the echo off, so it lands in no history and no file), the repository, the
`.claude/` tree and the mirror hook. It refuses a stdin that is not a terminal
and names the flags instead: every question it asks has one, and it performs
nothing itself, so an interactive run and a flag run are one code path.

Two rules that used to be prose are now the binary's: init refuses a linked
worktree and names the main checkout, and writing into an existing
`.claude/settings.json` is refused with the snippet printed rather than
reformatting a file the operator commits.

The scaffold version stamp went to its own `.kettle/scaffold.yaml` rather than
into `config.yaml`, because unknown keys there are a hard error and that file
may be committed and read by whatever build each machine has.

golang.org/x/term becomes a direct dependency; it was already in the tree
indirectly, so no module was added.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 16:17:24 +05:00

4.5 KiB

description, argument-hint, allowed-tools
description argument-hint allowed-tools
Make THIS directory a project that tracks issues — run `kettle init`, which creates the `.kettle/` marker every other command resolves the store from, migrates an older store in, gitignores it, and writes this command file and its skills. Carries the rules the binary cannot enforce — never an `--at` nobody named, never a migration clash resolved for them. [--at DIR] [--interactive] [--login NAME] [--repo owner/name] [--dry-run] Bash(kettle init:*), Bash(kettle config:*), Bash(git rev-parse:*)

/kettle:init — make this directory a project

Initializing is a statement, and the operator makes it: this directory is the project whose issues live in it. Nothing infers it — .git is in every clone, and a tool that inferred its root from one wrote other projects' issues into its own versioned cache. It is answered once, by a person, and every command downstream reads the answer instead of guessing.

This is a command, not a skill, and that is the point. It is invoked by an operator who typed it. There is no description: here for a model to match on, because "which directory is the project" is not a question a model gets to answer on its own.

The binary does the work and is idempotent. What this file carries is what it cannot decide.

Steps

  1. Run it, passing the operator's arguments through unchanged:

    kettle init $ARGUMENTS
    

    With no --at it initializes the current directory. Never supply an --at the operator did not name. Which directory is the project is the one question this command exists to have a person answer; picking a plausible one is the failure the marker replaces.

  2. Report what it printed, verbatim. already initialized — nothing to do is a success, not something to work around.

--login pins a login by name and --repo names the tracker repository; both are optional and both can be added later by running init again — it writes the config without disturbing settings it was not given. Neither is a credential: the tokens live in one file per machine, /kettle:auth.

--interactive is for a person at a terminal

It walks the operator through the login, the repository, this .claude/ tree and the AGENTS.md mirror hook, one question at a time.

Do not run it. It refuses a stdin that is not a terminal and says so, which is the right failure — but the right move is not to reach for it in the first place. Every question it asks has a flag: --login, --repo, --scaffold, --mirror-hook. Pass the flags the operator named and let the binary answer the rest.

When it stops

  • A linked worktree. The binary refuses, and it is right to: .kettle/ is gitignored, so a worktree has no marker by design and reaches the main checkout's store on its own — the walk crosses to it through the gitdir: in the .git file. A marker here would give one project two stores, and the second is deleted with the branch. What needs initializing, if anything, is the main checkout, and the refusal names it.
  • A name clash on the migration — the same file name on both sides. It exits having changed nothing and names the files. Report that. Do not move, delete, or merge either side: one of them may be an origin: local issue, which is the issue and the only copy of that work. The operator decides which survives.
  • A marker already exists above this directory. A second one gives that project a second store and the nearer one wins. Confirm with the operator before going ahead; usually they are standing in a subdirectory and there is nothing to do.

No kettle on PATH? command not found: kettle is the whole story — and it means the binary that wrote this file has since moved or been removed, because nothing else could have put it here. Stop and tell the operator to reinstall it: go install git.noodles.cam/claude-skills/marketplace/cli/cmd/kettle@latest, or cd cli && make install in a checkout of that repository (go.mod requires go 1.26).

After

  • /kettle:issue works now — offline, no login, no network.
  • /kettle:auth puts a token on this machine and pins the login this project runs under; needed only for the tracker side, /kettle:sync.
  • kettle config prints every path and setting this directory resolved to, and is the first thing to run when something looks like it landed in the wrong place. It also reports when this .claude/ tree was written by an older kettle than the one now installed.

The full flag table for init is the generated block in /kettle:project.