Kaelio/ktx-ai-data-agents-mcp-context-skills

ktx is a context layer that wraps your data and analytics stack and exposes it to AI agents through MCP, pre-built skills, and a memory system. The concrete problem it solves: raw MCP access lets an agent query your warehouse, but it does not give the agent the schema context, business logic, or query conventions it needs to return a correct answer rather than a plausible-sounding one. ktx closes that gap by bundling reusable skills and persistent memory alongside the MCP interface. The reason to look at this now is that Claude Code and Codex support is explicit, not bolted on — the repo is clearly built by someone who has run into the hallucinated-SQL problem in production. Reservation: this is early-stage infrastructure and the documentation reflects that — expect to read the source to fill gaps. Pricing is not a concern yet; it is fully open-source. -> Best for: AI engineer or technical PM building data-querying agents on top of Claude Code or Codex