メニュー
Classify a task and pick the right Claude model tier (haiku/sonnet/opus) before significant work. Use every time before spawning a sub-agent or starting substantial work, to right-size the model to the task.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Double-model verification gate before declaring a build/research/refactor "done". Runs a fresh Anthropic (Claude) review AND an independent OpenAI/Azure critic against the stated goal + criteria, then reconciles. Use before announcing completion of medium/major builds, refactors, major fixes, research deliverables, frontend work, or trade-decision rigor — and on demand via /verify. Skip for small Q&A and trivial edits.
Record and retrieve Architecture Decision Records — log a decision's context, options, rationale, and tradeoffs, then list/search them. Use when making a non-trivial design or architecture decision, choosing between tools/approaches, or asked 'why did we do it this way?'
Queue an external action (Telegram message, shell command, GitHub comment, PR merge, file write) for the user's explicit approval before it runs. Use when you want to take an outward-facing or irreversible action that needs sign-off first.
Run a structured multi-position debate over an analytical question, spawning sub-agents to argue each option and counter confirmation bias. Use when comparing options or tradeoffs — 'which is better', 'should we', 'what's the risk', 'evaluate', 'compare options'.
Ingest documents (URLs, local files, PDFs) into a searchable local knowledge base and query stored knowledge. Use when ingesting or storing documents, searching previously stored knowledge, or building research context.
Collect macro market and economic-calendar data (yfinance + FairEconomy API) into a SQLite DB and generate a formatted macro briefing for Telegram. Use when producing a macro/markets briefing or collecting macro market data.
| name | model-router |
| description | Classify a task and pick the right Claude model tier (haiku/sonnet/opus) before significant work. Use every time before spawning a sub-agent or starting substantial work, to right-size the model to the task. |
| category | meta |
| maturity | stable |
| tags | ["model-routing","tier-selection","cost","classifier","subagents"] |
Use EVERY TIME you are about to spawn a sub-agent or do significant work. Classify the task first, then use the right model tier.
node ${MODEL_ROUTER_HOME:-$HOME/projects/model-router}/classify.cjs "your task description here"
Or get JSON for scripting:
node ${MODEL_ROUTER_HOME:-$HOME/projects/model-router}/classify.cjs --json "task description"
sessions_spawn(task=..., model="<model from classifier>")
node ${MODEL_ROUTER_HOME:-$HOME/projects/model-router}/log-usage.cjs --task "..." --tier fast --model "..."
Use for: summarize, format, extract, classify, convert, translate, clean data, rename, sort, filter Cost: ~1× baseline When in doubt about simple tasks, start here.
Use for: explain, compare, draft, review, outline, moderate analysis, code with clear spec Cost: ~15× baseline
Use for: deep analysis, architecture decisions, complex tradeoffs, research synthesis, debates, long-form strategy Cost: ~75× baseline Reserve for tasks where quality clearly matters more than speed/cost.
Fast triggers: summarize, format, extract, classify, categorize, convert, translate, list, count, clean, parse, fix typos, rename, sort, filter, deduplicate, "quick", "simple", "brief", "just"
Powerful triggers: "analyze deeply", architecture, tradeoffs, debate, hypothesis, synthesize, "design system", "evaluate options", comprehensive, thorough, research, strategy, critique, multiple complex questions
node ${MODEL_ROUTER_HOME:-$HOME/projects/model-router}/usage-report.cjs
node ${MODEL_ROUTER_HOME:-$HOME/projects/model-router}/usage-report.cjs --days 30