English-language public beta · MCP-ready

Let agents think.
Give their work structure.

Run multiple agents in one local workspace. Give them reusable capabilities, connect tools near your data, and see what supports each result.

Get started free Read the docs

Built for teams worldwide. Bring your model, or connect an existing MCP client.

One local environment. Your Agents.

Workspace overview
Choose an Agent
Document assistant
Your operations Agent
Conversation

Bring the task.
Keep the context.

Talk, attach material and follow the work. Each Agent keeps its own conversations and model settings.

Install and sign in →
Execution and evidence

Worker
Tools run near your data.

Case
Inspect conditions and results.

Receipt
See the recorded outcome.

A map of the workbench. Console manages Agents, capabilities and permissions.

From conversation to governed work

Turn model output into controlled work

A model can suggest the next step. Production systems still need to know what evidence supports it, which policy applies, whether the action is allowed, and what actually happened.

Before execution

Rulith evaluates facts against the Agent’s installed Capabilities and Constitution. Missing evidence remains a visible gap instead of being guessed away.

After execution

Workers return structured results. Rulith records what ran, the evidence it produced, and whether the task satisfied its acceptance criteria.

How it works

Five steps from proposal to receipt

  1. Submit a real taskYour agent proposes facts, goals, or the next allowed action through MCP.
  2. Check the evidenceModel assertions stay distinct from source-attested data. Weak inputs cannot silently become strong conclusions.
  3. Apply Rules and ConstitutionThe installed Capabilities determine what can be concluded; the Constitution independently constrains what may proceed.
  4. Gate and execute the actionA worker performs the approved external call with bounded credentials and returns a structured result.
  5. Inspect the receiptThe Case keeps its conclusions, evidence chain, actions, outcomes, and acceptance result together.
Agent configuration

Install the Capability. Keep the Agent generic.

A Capability is one reviewable unit with four sections. Compatible Capabilities work together on the same Agent.

Vocabulary

The domain facts and shapes the Capability understands.

Rules

Criteria that derive conclusions, gaps, and the current frontier.

Actions

Governed transitions and typed requests for external work.

Sources

Named evidence origins and execution boundaries.

A Constitution constrains the Agent independently. Connections link that Agent to its Workers and external systems.

Start with something concrete

From a local document to a checked draft

The document assistant is one example of a capability: your Agent reads permitted material, proposes rules, runs local checks and presents a draft for review.

Available in Rulith

Build a capability from material

Attach a short text document, resolve questions, inspect examples and citations, then save a private draft. Publishing and installation stay explicit.

Walk through the example →

Your own workflow

Keep business logic in the capability

Define the facts, rules, actions and sources your work needs. Use the same execution and evidence machinery for every domain.

Create and configure a capability →

npm install --global rulithrulith startSetup guide →

Node.js 20+. The document assistant also requires Java 25. Remote model use can incur provider fees and sends authorized text to that provider.

Public beta

Build free during public beta

Platform payments are inactive during beta. Your model provider may charge separately. Any future platform pricing will be announced before activation.

Create an agent Explore the docs

Bring one agent workflow

Connect it, run real work, and tell us where the evidence, policy, or execution model breaks. Early feedback directly shapes the public API.

Get started free Send feedback

Developer questions and design-partner conversations: contact@rulith.com