Emerging
May 28, 20261
50%
Ktx: Open-source context layer enables AI agents to query data warehouses accurately
Ktx is an open-source context layer that automatically teaches AI agents how to query data warehouses accurately by ingesting company knowledge, mapping data infrastructure, and building semantic layers. Unlike traditional semantic layers that require manual maintenance, ktx automatically learns from wikis, detects database relationships, and flags contradictions while remaining read-only by design.
Quick Facts
Who
Kaelio (publisher)
What
Open-source executable context layer developed
When
2026-05-28
Where
Data warehouses
- Open-source executable context layer developed
- Automatically ingests and organizes company knowledge
- Maps data warehouses and detects joinable columns
- Builds semantic layers with automatic join graph resolution
- Serves agents through CLI and MCP tools
Ktx is a new open-source tool that provides an executable context layer designed to improve how AI agents interact with data warehouses. The software addresses a fundamental limitation of general-purpose AI agents: their tendency to re-explore data warehouses on every query, invent their own metric definitions, and produce results that diverge from approved business standards.
The tool works by automatically building and maintaining context from multiple sources within an organization. Ktx ingests wiki content from platforms like Notion, organizes business knowledge, detects joinable columns in databases, and flags contradictions across sources. It combines raw database tables with high-level metrics through an automated join graph that resolves common data modeling problems known as chasm and fan traps. This enables AI agents to fetch metrics declaratively rather than rewriting SQL queries repeatedly.
Ktx differentiates itself from traditional semantic layers, which typically require constant manual maintenance and do not absorb broader company knowledge. The tool operates as a read-only system by design and integrates with multiple data platforms including PostgreSQL, Snowflake, BigQuery, ClickHouse, MySQL, SQL Server, and SQLite. It also connects with popular analytics and workflow tools such as dbt, Looker, Metabase, and MetricFlow. The system exposes both command-line and MCP (Model Context Protocol) tools that enable agents to search across business knowledge and semantic layer definitions.
Installation is straightforward through npm, with setup involving local project configuration, provider setup, and context building. The software is designed for organizations that have scattered business knowledge across multiple platforms and want AI agents like Claude Code, Codex, Cursor, or OpenCode to query warehouses using approved metric definitions. Ktx can be used with personal LLM API keys or through a Claude Pro/Max subscription, with no additional usage billing from the tool itself.
Topics
Why This Matters
Ktx addresses a critical pain point for organizations deploying AI agents in data environments: ensuring consistency, accuracy, and compliance with business standards. By automating context layer management and eliminating repetitive SQL rewrites, it reduces maintenance overhead while enabling AI systems to generate reliable metrics that align with approved definitions. This is particularly valuable for teams scaling AI-driven analytics across distributed knowledge sources.
Timeline & Sources
May 28, 2026
WireKtx open-source context layer announced on Hacker News