What is Ktx used for?

Ktx is an open-source context layer that helps AI agents such as Claude Code and Codex query data warehouses accurately using learned metrics, joins and company knowledge. It is a good fit for letting AI agents answer metric questions from a warehouse, replacing a hand-maintained semantic layer and keeping agent answers aligned with approved definitions.

Yes. Ktx is open source under the Apache-2.0 licence. Its source code is on GitHub at Kaelio/ktx and is written mainly in TypeScript.

Yes. Ktx is open source, so the software itself is free to use.

Yes. Ktx can be self-hosted on your own server or infrastructure.

What are some alternatives to Ktx?

Similar open-source tools in the AI Infrastructure category include Observal, OpenConnector and Inkeep Agents. SaaS products in the same category include AgentGov, Amazon Bedrock and AnyRouter.

Is Ktx actively maintained?

Yes. The most recent commit to Ktx was on 11 September 2026, and the latest release is v0.16.0, published on 3 July 2026. The project has 1.6k stars on GitHub.

Ktx: open-source AI infrastructure

About Ktx

ktx is a context layer for data and analytics agents, built and maintained by Kaelio. General-purpose coding agents tend to re-explore a warehouse on every question and invent their own metric logic, which produces numbers that do not match approved definitions. ktx gives them a maintained body of context so answers rely on agreed definitions instead.

It works in several steps. It ingests wiki content, organizes it, removes duplicates and flags contradictions for human review. It samples tables, captures metadata and usage patterns, detects joinable columns and annotates sources. It then builds a semantic layer with a join graph that resolves fan and chasm traps, so agents can request metrics declaratively. Finally, it exposes CLI and MCP tools with combined full-text and semantic search.

The project is written in TypeScript under the Apache-2.0 licence. You run ktx with your own LLM API keys or a local agent sign-in such as Claude Code or Codex, and the README says ktx adds no extra usage billing. It suits data and analytics engineering teams that want AI agents to answer business questions from a warehouse without hand-maintaining a traditional semantic layer.

Key features

Automatic mapping of warehouse tables and joins
Semantic layer with approved metric definitions
Ingestion and deduplication of company wiki content
CLI and MCP tools for agents
Combined full-text and semantic search
Works with Claude Code and Codex

Good fit for

→Letting AI agents answer metric questions from a warehouse
→Replacing a hand-maintained semantic layer
→Keeping agent answers aligned with approved definitions

Built with: TypeScript
Tags: data-agents; semantic-layer; mcp; analytics; claude-code; context-layer; text-to-sql; typescript; ai-agents

Ktx: questions and answers

What is Ktx used for?: Ktx is an open-source context layer that helps AI agents such as Claude Code and Codex query data warehouses accurately using learned metrics, joins and company knowledge. It is a good fit for letting AI agents answer metric questions from a warehouse, replacing a hand-maintained semantic layer and keeping agent answers aligned with approved definitions.
Is Ktx open source?: Yes. Ktx is open source under the Apache-2.0 licence. Its source code is on GitHub at Kaelio/ktx and is written mainly in TypeScript.
Is Ktx free?: Yes. Ktx is open source, so the software itself is free to use.
Can I self-host Ktx?: Yes. Ktx can be self-hosted on your own server or infrastructure.
What are some alternatives to Ktx?: Similar open-source tools in the AI Infrastructure category include Observal, OpenConnector and Inkeep Agents. SaaS products in the same category include AgentGov, Amazon Bedrock and AnyRouter.
Is Ktx actively maintained?: Yes. The most recent commit to Ktx was on 11 September 2026, and the latest release is v0.16.0, published on 3 July 2026. The project has 1.6k stars on GitHub.