Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Manual publish after 1.0.9 release merge
What This Skill Does
Runs multi-LLM decision workflows using the @consensus-tools/consensus-tools engine. Agents post jobs, submit artifacts, vote, and resolve outcomes through configurable consensus policies. Operates locally by default; a hosted board mode is available when explicitly configured.
Supports multiple resolution policies (approval voting, confidence ranking, first submission) and runs entirely local without requiring a hosted service, making it usable offline or in restricted environments.
When to use it
- Picking the best agent-generated code fix by approval vote
- Settling conflicting outputs from parallel LLM calls before acting on the result
- Gating an automated pipeline step until a quorum of agents agrees
- Running a local multi-agent decision loop during development without a remote server
- Resolving disagreements between agents on which API response to trust
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Have agents vote on the best deployment strategy and return the winner
- 1Post a job in VOTING mode with APPROVAL_VOTE policy using consensus-tools_post_job
- 2Each agent submits a deployment strategy artifact with a confidence score via consensus-tools_submit
- 3Voters cast yes/no votes on each submission using consensus-tools_vote
- 4Resolve the job with openclaw consensus resolve <jobId>
- 5Fetch the winning artifact with consensus-tools_status or openclaw consensus result get <jobId>
The submission with the highest approval score is returned as the trusted output
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.