multi-agent-orchestration
par dabruhce
Productivity & TasksRésumé
Multi-agent task delegation and process orchestration with audit logging and agent learning capabilities.
Description
Ce que fait ce skill
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.
Quand l'utiliser
- 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
Exemple de workflow
User asks: validate a business idea for an AI meal planning app for parents
- 1
Run
node scripts/colony.mjs process validate-idea --context "AI meal planning for busy parents" - 2
Colony spawns muse for brainstorm stage, then scout for research, then forecast for analysis
- 3
Process pauses at checkpoint after analyze stage and sends a notification
- 4
User runs
node scripts/colony.mjs approve <run-id>to continue - 5
forge and ledger run in parallel to produce spec and cost estimate, outputting business-case.md
business-case.md with market research, competitive analysis, product spec, and cost estimate
Installation
clawhub install dabruhce/multi-agent-orchestrationExécute la commande CLI, ou colle l'instruction dans Claude Code, Cursor, Codex ou l'agent de code que tu utilises.