78 added, 129 removed. Audit B to A.
---
name: setup
description: >-
- Walks a non-technical user through connecting every judge model this plugin
- can use — Codex, OpenCode, OpenRouter, Gemini — one at a time, in plain
- words. Use for "multi setup", "set up multi", "configure judges/reviewers/
- backends", "onboarding", or whenever a probe shows missing backends and the
- user needs to know what to do about it.
- allowed-tools: Bash, Read, Grep, Glob
+ Walks a user through connecting the judge models this plugin can use — Codex,
+ OpenCode, OpenRouter (or any compatible endpoint: 9router, z.ai, Moonshot),
+ Gemini — and through picking models, including "research current models for
+ me". Use for "multi setup", "set up multi", "configure judges/reviewers/
+ backends", "connect 9router", "change/pick models", "onboarding", or whenever
+ a probe shows missing backends and the user needs to know what to do about it.
+ allowed-tools: Bash, Read, Grep, Glob, WebSearch, WebFetch
argument-hint: "[nothing needed — just run it]"
---
# Connect the judges
!`sh -c 'for p in "$CLAUDE_PLUGIN_ROOT/scripts" "$HOME/.claude/skills/multi/scripts" "./.claude/skills/multi/scripts"; do [ -x "$p/probe.sh" ] && { "$p/probe.sh"; echo "scripts-dir: $p"; exit 0; }; done; echo "probe: NOT FOUND — locate scripts/probe.sh in this plugin and run it yourself"'`
`$SCRIPTS` is whatever the probe printed as `scripts-dir:`. The probe already
- ran above — read it, do not run it again.
+ ran above — read it, do not run it again. It detected everything detectable:
+ which CLIs are installed, who is logged in, which keys exist, whether the
+ OpenCode paid channel is available, and what other AI CLIs live on this
+ machine. **Never ask the user about anything the probe already answered.**
The user here is not assumed technical. No jargon, no walls of text, one step
at a time, and never make them feel behind for not having done this already.
- ## What this plugin actually is, in plain words
-
- Say this first, briefly, before touching any config:
-
- This plugin asks several different AI models to look at the same thing
- instead of just one. When they agree, that is a real signal. When they
- disagree, that disagreement is the whole point — one model alone can't tell
- you it might be wrong, three models can.
-
- The lineup:
-
- - **Claude** — already here, always available, nothing to configure.
- - **GPT**, through the Codex CLI — needs a ChatGPT account.
- - **DeepSeek**, through OpenCode — free, works out of the box once installed.
- - **Gemini**, through its own CLI — needs a free API key.
- - **OpenRouter** — the bonus one. A single key unlocks many other model
- families at once. Free models by default, or any specific model with
- `--backend name:model` (e.g. `--backend openrouter:z-ai/glm-5.2:free`).
-
- ## Fix one thing at a time
-
- Look at the probe output above. For each backend that is missing or not
- configured, offer the fix — **one at a time, not all four in a list.** Start
- with the most valuable one that's missing. Codex comes first if it's missing:
- `/multi:code-review` refuses to run at all without it, the others just get
- weaker.
+ ## First contact: what this is, in three sentences
- Order to work through, skipping each install or login fix already `OK`. The
- OpenCode model-choice check below always runs, even when OpenCode is `OK`:
+ Only when little or nothing is connected yet (Claude alone, or one backend).
+ Returning users with a working lineup skip straight to fixing gaps.
- 1. **Codex missing or not logged in**
- ```
- codex login
- ```
- Needs a ChatGPT account. This opens a browser to sign in — nothing to type
- here.
+ > This plugin asks several different AI models to look at the same thing
+ > instead of just one — when they agree it's a real signal, and when they
+ > disagree, that disagreement is exactly what one model alone can never show
+ > you. You'll use it mostly as `/multi:code-review` (several models review
+ > your changes before a PR), `/multi:ask` (one question, several independent
+ > answers), and `/multi:check-if-done` (is this actually finished?).
+ > Claude is already connected; each backend below adds an independent
+ > reviewer, most at no extra cost.
- 2. **OpenCode**
- If it is missing:
- ```
- curl -fsSL https://opencode.ai/install | bash
- ```
- (or `npm install -g opencode-ai` if they'd rather use npm). Nothing to pay
- for or sign up to — it works with free models right out of the box.
+ ## One question, only about what the probe can't see
- Once OpenCode is checked — whether it was already `OK` or was just
- installed — check its model choice. The config is
- `${XDG_CONFIG_HOME:-$HOME/.config}/multi/models`, unless
- `MULTI_MODELS_CONFIG` overrides it. It is a hand-edited file of
- `key: value1 value2 ...` lines with `#` comments. Today the only key is
- `opencode:`; its first model is primary and the rest are fallbacks.
+ The probe knows what's installed. It cannot know what accounts the user has
+ or what they're willing to pay. Ask **one** question covering only the
+ missing backends, shaped like this (adjust to the actual gaps):
- Determine the effective primary model in this exact order: use
- `MULTI_OPENCODE_MODEL` when it is set; otherwise use the first model on the
- config's `opencode:` line when present; otherwise use the auto-detected
- primary reported by the probe above. If the config has an `opencode:` line,
- still say which models are pinned, in order, but do not use the line's mere
- presence or absence to decide whether the active channel is free.
+ > To pick what to connect, tell me:
+ > - Do you have a ChatGPT subscription? (unlocks GPT as a reviewer, no extra cost)
+ > - Are you okay creating one free account + API key? (unlocks many model
+ > families at once via OpenRouter or a compatible service)
+ > - Or keep it strictly free-and-local for now? (OpenCode's free models)
- Decide the warning from the effective primary model's provider prefix:
- `opencode/*` is the free channel and `opencode-go/*` is the paid Go
- subscription channel. Warn **only** when the effective primary is an
- `opencode/*` free model. Thus an environment override takes precedence over
- a configured model, and a configured free primary still gets the warning.
- For a free effective primary, say this **loudly and plainly**:
+ Then recommend an order — the fewest steps to reach **at least one non-Claude
+ reviewer** (without one, `/multi:code-review` refuses to run; with one it
+ works, and each further backend makes it stronger). Typical value order:
+ Codex if they have the subscription, otherwise an OpenRouter-style key,
+ OpenCode as the free floor, Gemini as a free extra.
- > **Important: you're using free models. They are weaker and less stable. A
- > review may come back thin or even empty, so reviewing on free is a real
- > quality compromise.**
+ ## Connect each chosen backend
- Then ask an explicit yes/no question:
+ Work through the user's picks **one at a time** — give one step, wait for
+ them to do it, then the next. Never paste four install blocks in one message.
- > Want me to look up the current models, their prices, and how much usage
- > you'd get (how much you'd pay) — and help you pick a good one + fallbacks?
+ For each backend, read its reference file first and follow it:
- If **no**, acknowledge that keeping free is a deliberate choice and move
- on. If **yes**:
+ - **Codex (GPT)** — `references/codex.md`
+ - **OpenCode** — `references/opencode.md`
+ - **OpenRouter / 9router / z.ai / any compatible endpoint** — `references/openrouter.md`
+ - **Gemini** — `references/gemini.md`
- - Run both `opencode --pure models` and
- `opencode --pure models --verbose`; `--pure` is required so plugin noise
- does not pollute the output, and the verbose listing includes costs.
- - Use web search for unfamiliar model names: your own model knowledge is
- stale and newer models will not be in it. Check what they actually are,
- their quality and benchmarks, and current pricing.
- - Propose one primary model and an ordered fallback chain.
- - After the user chooses, create the config directory if needed and write
- `opencode: <primary> <fallback1> <fallback2> ...`, preserving every other
- line already in the file.
+ If the probe printed a `models-config: LEGACY PATH` line, offer the one-line
+ `mv` it shows — everything of this plugin lives under `~/.claude/multi/`.
- Remind them that this is deliberately hand-editable: they can change that
- one `opencode:` line themselves at any time, with no agent needed.
+ If the probe listed `other-ai-clis`, you may mention them in one sentence as
+ detected-but-not-yet-supported. Do not improvise support for them.
- 3. **OpenRouter not configured**
- Key comes from [openrouter.ai](https://openrouter.ai) — sign up, create a
- key. There's a free tier, no card needed to get one key working. Then they
- run this **themselves, in their own terminal** — it asks for the key and
- does not echo it:
- ```
- $SCRIPTS/setup.sh set OPENROUTER_API_KEY
- ```
+ ## Picking models
- 4. **Gemini not configured**
- Key is free, from [aistudio.google.com](https://aistudio.google.com). The
- second line is theirs to run in their own terminal, same as above:
- ```
- npm i -g @google/gemini-cli
- $SCRIPTS/setup.sh set GEMINI_API_KEY
- ```
+ When a backend is connected but the model choice is the question (free vs
+ paid, which pin, which fallbacks) — or the user asks "which models should I
+ use" — offer the fork:
- Give one fix, wait for them to do it or ask a question, then move to the
- next. Do not paste all four blocks in one message — that's the thing this
- skill exists to avoid.
+ 1. **Sane default** — keep what the probe picked; say in one line what that
+ is and what it costs (usually: free, weaker, can flake).
+ 2. **"Research it for me"** — read `references/model-research.md` and follow
+ it: live catalogue + web search (your own model knowledge is stale),
+ propose a pick, write the config only after the user approves.
- ## Never ask for a key in chat, and never run the command yourself
+ ## Keys never touch this chat
- Not a preference — a key that reaches this conversation is a leaked key: it is
- in the transcript, and if it also reached a command line it is in their shell
- history and was visible in `ps` to everyone on the machine. So `setup.sh set
- <NAME>` is given **without** a value and run **by them**, in their own
- terminal: it prompts, reads the key without echoing it, and writes it out of
- sight. If a key does land in the chat anyway, say plainly that it should be
- rotated at the provider — a leaked key is not un-leaked by deleting a message.
+ Not a preference — a key that reaches this conversation is a leaked key: it
+ is in the transcript, and if it also reached a command line it is in shell
+ history and was visible in `ps`. So `setup.sh set <NAME>` is given **without**
+ a value and run **by the user, in their own terminal**: it prompts, reads the
+ key without echoing, and writes it out of sight. If a key lands in the chat
+ anyway, say plainly it should be rotated at the provider — deleting a message
+ does not un-leak it.
- Keys live outside this plugin's folder, in `~/.claude/multi/providers.env`,
- which `scripts/setup.sh` sets to permission `600` — only their own user can
- read it. On Windows/MSYS and on some mounts (exFAT, some NTFS) `chmod` is
- accepted and does nothing; the script checks afterwards and prints a warning
- when that happened, so if you see no warning the file really is owner-only,
- and if you see one, repeat it to them rather than restating the `600` line.
+ Keys live in `~/.claude/multi/providers.env`, permission `600` (owner-only).
+ On Windows/MSYS and some mounts `chmod` silently does nothing; `setup.sh`
+ checks afterwards and warns when that happened — if it warned, repeat the
+ warning to the user instead of claiming the file is protected.
## Finish: show where things stand
Run:
```
$SCRIPTS/setup.sh status
```
- Report it as one short block — who's connected, and what each missing one
- would still add:
+ This one actually verifies keys over the network (the probe above does not).
+ Report one short block — who's connected, and what each missing one would add:
```
- Connected: Claude, Codex
- Not yet: OpenCode (free third opinion), OpenRouter (bonus — many models via one key), Gemini (free fourth opinion)
+ Connected: Claude, Codex, OpenRouter (endpoint: 9router)
+ Not yet: OpenCode (free extra reviewer), Gemini (free extra reviewer)
- With just Claude: /multi:code-review refuses to run (needs Codex at minimum).
- /multi:ask and /multi:adhd still work, just with fewer opinions.
+ /multi:code-review works — you have non-Claude reviewers. Each missing
+ backend is one more independent opinion, most at no extra cost.
```
- Be honest about degradation, not alarming about it. Missing backends are a
- normal starting state, not a broken install.
-
- Then stop. Don't chain into another skill or start reviewing anything —
- this skill's only job is getting the judges connected.
+ Be honest about degradation, not alarming: missing backends are a normal
+ starting state, not a broken install. Then stop — this skill's only job is
+ getting the judges connected, not running a review.