Lewati ke konten
TakoAPI
Kembali ke skill

multi-agent-orchestration

oleh dabruhce

Productivity & Tasks
812Unduhan dilaporkan oleh clawskills.sh
4Tayangan
0 sukaMasuk untuk menyukai skill ini
—Belum ada penilaian
Penilaian Anda

Ringkasan

Multi-agent task delegation and process orchestration with audit logging and agent learning capabilities.

Deskripsi

Apa yang Dilakukan Skill Ini

Colony is a multi-agent orchestration system with 12 specialized agents (researcher, coder, writer, QA, etc.) that can be dispatched individually or chained into multi-stage processes. It includes an audit log, per-agent persistent memory, shared cross-agent learnings, and checkpoint-based human approval gates.

Having named specialist agents with persistent memory and cross-agent shared learnings means tasks accumulate context over time rather than starting cold each run.

Kapan Menggunakannya

  • Validating a startup idea end-to-end with research, analysis, and cost estimation
  • Running a content pipeline from research draft through social promotion
  • Triaging and fixing a bug across reproduce, fix, test, and deploy stages
  • Delegating a quick competitive research query without specifying which agent
  • Reviewing agent performance stats and failure patterns after a sprint

Contoh Alur Kerja

INPUT

User asks: validate a business idea for an AI meal planning app for parents

AGENT
  1. 1

    Run node scripts/colony.mjs process validate-idea --context "AI meal planning for busy parents"

  2. 2

    Colony spawns muse for brainstorm stage, then scout for research, then forecast for analysis

  3. 3

    Process pauses at checkpoint after analyze stage and sends a notification

  4. 4

    User runs node scripts/colony.mjs approve <run-id> to continue

  5. 5

    forge and ledger run in parallel to produce spec and cost estimate, outputting business-case.md

OUTPUT

business-case.md with market research, competitive analysis, product spec, and cost estimate

Instalasi

clawhub install dabruhce/multi-agent-orchestration

Jalankan perintah CLI-nya, atau tempel prompt ke Claude Code, Cursor, Codex, atau coding agent apa pun yang Anda pakai.

Tautan