๋ณธ๋ฌธ์œผ๋กœ ๊ฑด๋„ˆ๋›ฐ๊ธฐ
TakoAPI
์Šคํ‚ฌ๋กœ ๋Œ์•„๊ฐ€๊ธฐ
โ€”์•„์ง ํ‰๊ฐ€ ์—†์Œ
๋‚ด ํ‰๊ฐ€

์š”์•ฝ

# ๐Ÿง  Skill Router (Skill Orchestrator)

์„ค๋ช…

์ด ์Šคํ‚ฌ์ด ํ•˜๋Š” ์ผ

Acts as a governance layer above other installed skills. It inventories available skills, scores each candidate against the task using a weighted model, applies safety gates, and executes the best option with full user visibility. An optional multi-agent brainstorming module adds business/strategy and market/execution agents for structured idea generation rounds.

The deterministic 0-100 scoring model with visible per-component breakdowns replaces guessing which skill fits a task.

์‚ฌ์šฉ ์‹œ์ 

  • Automatically selecting the right skill for an ambiguous or complex task
  • Previewing what would execute with a dry-run before committing
  • Comparing scored skill candidates side-by-side before choosing one
  • Running structured multi-agent brainstorming sessions for business strategy
  • Auditing installed skills, their risk levels, and required permissions

์˜ˆ์‹œ ์›Œํฌํ”Œ๋กœ

์ž…๋ ฅ

User asks: route this task automatically โ€” summarize last week's sales data

์—์ด์ „ํŠธ
  1. 1

    Normalize the request into goal, constraints, urgency, and risk profile

  2. 2

    Inventory installed skills via platform API or directory scan and manifest parsing

  3. 3

    Score each candidate on task relevance, environment fit, permission fit, latency, risk alignment, and local history

  4. 4

    Apply safety gates and request confirmation if risk level is HIGH or credentials are involved

  5. 5

    Execute the top-scoring skill and emit a structured Skill Router Report

์ถœ๋ ฅ

A structured report listing scored candidates, the selected skill, safety check result, and execution outcome

์„ค์น˜

clawhub install myboxstorage/brainstorming-studio

CLI ๋ช…๋ น์„ ์‹คํ–‰ํ•˜๊ฑฐ๋‚˜, ํ”„๋กฌํ”„ํŠธ๋ฅผ Claude Code, Cursor, Codex ๋“ฑ ์‚ฌ์šฉํ•˜๋Š” ์ฝ”๋”ฉ ์—์ด์ „ํŠธ์— ๋ถ™์—ฌ๋„ฃ์œผ์„ธ์š”.

๋งํฌ