AiMatch

Kyno

Kyno lets AI agent operators publish versioned missions and principles that agents retrieve through MCP before a workflow step. Teams can update direction during a run and review which version was supplied to each step.

One of 576 tools in AI Agents

Visit Kyno Compare alternatives Own this tool? Claim it
Kyno screenshot

Who it's for

  • AI agent developers using CrewAI or LangGraph who need agents to retrieve current mission and principles at step boundaries.
  • Operators managing agent workflows who need to change priorities while a run continues and review the direction supplied to steps.
  • Developers building SDK or custom adapters who want to connect an agent workflow to Kyno's direction store.

How it fits your workflow

An operator creates a constitution containing a mission and ordered principles, then publishes it as a version. The page's example uses a YAML file and applies it with a note.

An adapter pulls the current direction through MCP before an agent's next step. Kyno records the supplied version, which can later be retrieved when reviewing the run; the page says failed pulls follow the configured policy.

Pricing

The vendor doesn't publish prices. Check with Kyno directly.

Prices checked on Oct 2, 2026 from the vendor's site. They can change; confirm before you buy.

Key features

Versioned constitutions
A constitution is a named mission and ordered principles with a current version and immutable history. Operators can publish a change while a workflow continues.
Direction retrieval at step boundaries
Adapters pull current direction before work proceeds. Successful pulls return the current direction; failed pulls follow the configured policy.
Direction version attribution
Kyno records the direction version supplied to each step so that the mission and principles used for a run can be retrieved during review.
Multiple transport and adapter options
The page describes MCP transport, a SQLite and stdio starting setup, and adding HTTP or a custom adapter as workflow needs require.

Works with

  • MCP
  • CrewAI
  • LangGraph
  • SDK / custom adapter
  • HTTP

Limitations to know

  • The stated requirement is Python 3.11 or later.

Getting started

pip install kyno

Python 3.11+

  1. Install Kyno from PyPI and create a project with `kyno new acme`.
  2. Initialize the database with `kyno db init`; the page recommends starting with SQLite and stdio.
  3. Create a constitution YAML file and apply it with `kyno apply constitution.yaml --note "initial constitution"`.
  4. Run `kyno serve --transport stdio`; add HTTP or a custom adapter when the workflow needs it.