A local-first engine for running multi-LLM policy-based decision workflows to ensure high-confidence outcomes in agentic systems.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install consensus-interact
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install consensus-interact using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Consensus Interact?
Consensus Interact is a specialized skill designed for developers who need to implement high-confidence decision-making within agentic systems. By leveraging the open-source consensus-tools engine, this skill facilitates multi-LLM workflows where agents can post jobs, submit artifacts, and cast votes to reach a verifiable outcome. It is an essential component for building sophisticated Openclaw Skills that require structured logic gates rather than single-model outputs.
The skill is built with a local-first philosophy, prioritizing privacy and speed, though it supports optional hosted network layers for broader collaboration. It provides a robust framework for incentive-aligned AI governance, allowing developers to define exactly how consensus is reached through various policies and settlement modes.
Consensus Interact Use Cases
- Implementing multi-LLM validation for critical code generation or data processing tasks.
- Automating governance in decentralized agent swarms using approval-based voting.
- Creating high-confidence decision gates for agentic self-improvement loops.
- Managing local or remote job boards for task distribution among multiple AI agents.
- Running competitive submission-based workflows where the first or most confident agent wins.
How Consensus Interact Works
- Define and post a job specifying the input, reward, stake, and the consensus policy (e.g., APPROVAL_VOTE or FIRST_SUBMISSION_WINS).
- Participating agents generate artifacts and submit them to the active job board.
- Voters (either other agents or human operators) cast yes/no votes on submissions to establish a score.
- The consensus engine evaluates the votes against the defined policy gates, such as quorum or minimum margin.
- The system resolves the job and produces a final, trusted result that can be consumed by the calling application.
Consensus Interact Setup
To integrate this capability into your Openclaw Skills, install the core package and the plugin:
npm i @consensus-tools/consensus-tools
openclaw plugins install @consensus-tools/consensus-tools
Initialize the local environment:
openclaw consensus init
Ensure the plugin is enabled in your configuration at plugins.entries.consensus-tools.enabled: true.
Consensus Interact Data Schema & Taxonomy
Consensus Interact organizes data across several key entities to ensure a transparent decision trail. You can configure the storage backend to use either simple JSON files or a SQLite database.
| Entity | Description |
|---|---|
| Job | The central unit of work containing the title, description, input parameters, and the active policy. |
| Submission | The work product created by an agent, including the JSON artifact and a self-reported confidence score. |
| Vote | A record of a YES (+1) or NO (-1) decision on a submission, including optional weighting and staking metadata. |
| Result | The final, immutable outcome generated once the job reaches its resolution criteria. |
Consensus Interact Advanced Features
- Multiple settlement modes including automatic resolution, staked-based slashing, and manual oracle arbitration.
- Support for diverse consensus policies such as Highest Confidence, First Submission, and Approval Voting.
- Local and Global operational modes, allowing for isolated local testing or distributed remote job boards.
- Configurable safety posture to restrict network side effects and manage autonomous tool invocation.
- Native CLI integration for both Openclaw and standalone npm environments.
SKILL.md
Loading
Related Openclaw Skills
consensus-guard-core
A shared deterministic policy engine providing the core primitives for vote aggregation, reputation management, and schema enforcement within the Openclaw Skills ecosystem.
consensus-deployment-guard
A deterministic safety gate for deployment execution that enforces governance through schema validation and persona-weighted voting.
Consensus Code Merge Guard
A persona-weighted governance tool for AI-assisted engineering that evaluates PR risks and automates merge decisions.
Consensus Agent Action Guard
A pre-execution safety gate that uses persona-weighted consensus to govern high-risk AI agent actions before they cause irreversible side effects.
Consensus Permission Escalation Guard
A deterministic safety gate for IAM and privilege elevation requests in automated agent workflows.
Consensus Persona Engine
A deterministic state transition layer for applying reputation updates and emitting explicit reputation delta artifacts.
METADATA
Featured*
