Task External Models

What it does

When your agent uses it

• Confused about how to run external models
• Claudish with Bash
• External model in /team
• How to specify external model

Example prompts

• “claudish with Bash”
• “external model in /team”
• “how to specify external model”
• “/task-external-models”

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

1. 1Detect context from task keywords (debug/research/coding/review)
2. 2If contextPreferences[context] has models → USE THEM (no asking)
3. 3If empty (first time for context) → ASK user → SAVE to that context
4. 4User says "use different models" → ASK and UPDATE

What it can do on your machine

Read from SKILL.md and the folder at commit 6097ad4. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  No scripts in the folder and no shell commands in SKILL.md (its code samples are bash and javascript).

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Task External Models loads about 1.4k tokens when it runs. Until then it costs about 103 tokens; SKILL.md has 304 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~103

When it runs · the whole SKILL.md, loaded when a task matches

~1.4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from MadAppGang/claude-code at commit 6097ad4, republished under its MIT licence (© MadAppGang). 304 words, ~1,365 tokens.

name

task-external-models

description

Quick-reference for using external AI models in orchestration workflows. External models are invoked via Bash+claudish CLI (deterministic, 100% reliable). Use when confused about how to run external models, "claudish with Bash", "external model in /team", or "how to specify external model". Trigger keywords - "external model", "claudish", "Bash claudish", "external LLM", "model parameter".

version

2.0.0

tags

external-model, quick-reference, bash, claudish, agent-cli

keywords

external model, grok, gemini, gpt-5, minimax, claudish, bash, external LLM, cli

plugin

multimodel

updated

2026-02-11

cat .claude/multimodel-team.json 2>/dev/null

claudish --model {MODEL_ID} --stdin --quiet < prompt.md > result.md

# Pattern
claudish --model {MODEL_ID} --stdin --quiet < prompt.md > result.md 2>stderr.log; echo $? > result.exit

# Examples
claudish --model x-ai/grok-code-fast-1 --stdin --quiet < task.md > grok.md 2>grok-err.log; echo $? > grok.exit
claudish --model google/gemini-3-pro-preview --stdin --quiet < task.md > gemini.md 2>gemini-err.log; echo $? > gemini.exit
claudish --model openai/gpt-5.2-codex --stdin --quiet < task.md > gpt5.md 2>gpt5-err.log; echo $? > gpt5.exit

claudish [options]

--model <id>         AI model to use (e.g., x-ai/grok-code-fast-1)
--stdin              Read prompt from stdin
--quiet              Minimal output

// Internal model via Task
Task({
  subagent_type: "dev:researcher",
  description: "Internal Claude vote",
  run_in_background: true,
  prompt: "{VOTE_PROMPT}\n\nWrite to: {SESSION_DIR}/internal-result.md"
})

// External models via Bash+claudish (all in same message)
Bash({
  command: "claudish --model x-ai/grok-code-fast-1 --stdin --quiet < {SESSION_DIR}/vote-prompt.md > {SESSION_DIR}/grok-result.md 2>{SESSION_DIR}/grok-stderr.log; echo $? > {SESSION_DIR}/grok.exit",
  run_in_background: true
})

Bash({
  command: "claudish --model google/gemini-3-pro-preview --stdin --quiet < {SESSION_DIR}/vote-prompt.md > {SESSION_DIR}/gemini-result.md 2>{SESSION_DIR}/gemini-stderr.log; echo $? > {SESSION_DIR}/gemini.exit",
  run_in_background: true
})

# Get current available models
claudish --top-models    # Best value paid models
claudish --free          # Free models

# Example model IDs (verify with commands above)
x-ai/grok-code-fast-1       # Grok (fast coding)
minimax/minimax-m2.5        # MiniMax M2.5
google/gemini-3-pro-preview # Gemini Pro
openai/gpt-5.2-codex        # GPT-5.2 Codex
z-ai/glm-4.7                # GLM 4.7
deepseek/deepseek-v3.2      # DeepSeek v3.2

For each external model result:
  ☐ Exit code is 0
  ☐ Result file exists and has >50 bytes
  ☐ Response contains substantive analysis (not just acknowledgment)
  ☐ No error messages in stderr log

© MadAppGang, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in plugins/multimodel/skills/task-external-models of MadAppGang/claude-code.

Open the folder on GitHubat commit 6097ad4