Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Docs: added complexity explanation section — why the skill is large, why scanners flag it, what each component does. Helps users and automated scanners understand scope vs. malice.
What This Skill Does
Patterns and production-ready reference libraries for orchestrating multi-agent systems in OpenClaw. Covers security isolation, parallel research with multi-lens synthesis, phased implementation pipelines, peer validation, cost estimation, and intent drift detection. Each pattern ships with a JS library and a spawn template.
Instead of designing isolation, cost-gating, and drift-detection patterns from scratch, you get tested JS libraries and spawn templates with documented security tradeoffs already worked out.
When to use it
- Isolating an untrusted third-party API behind a sandboxed security proxy
- Running parallel optimist/pessimist/pragmatist researchers on a technology adoption decision
- Breaking a feature build into architect, coder, and reviewer phases with per-phase cost caps
- Pre-estimating subagent spend and blocking spawns that exceed a budget threshold
- Detecting when an agent's outputs have drifted from its stated purpose over time
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Should we migrate our backend to FrameworkX?
- 1Determine task complexity is high and select the research team pattern
- 2Spawn three parallel researchers with distinct lenses: optimist (best-case analysis), pessimist (risks and failure modes), pragmatist (real-world adoption data)
- 3Each researcher independently gathers and validates sources, posting findings to a shared append-only chatroom
- 4UnifierAgent reads all three lens outputs and synthesizes a balanced decision tree with tradeoffs and adoption criteria
Structured recommendation covering upside potential, known risks, current ecosystem maturity, and a go/no-go criteria checklist