Skip to content

Skills for Claude Code and Codex

This page introduces KDB.AI skills for Claude Code and Codex, and explains how to install and use them.

Claude Code and OpenAI Codex both support Skills — packaged, versioned instructions the agent loads automatically when a conversation matches the task they cover. kdbai-knowledge is a plugin that bundles KDB.AI skills for schema design, search, and sizing. Install the plugin once, then ask KDB.AI-related questions in natural language from your editor or terminal.

What it covers

  • Schema and search guidance – table and index design, hybrid search, filters, reranking, and common integration patterns for the KDB.AI Python and q APIs.
  • Sizing and capacity planning – turns a dataset shape and workload into an index shortlist and a RAM/disk/GPU VRAM estimate, following the same methodology as the deployment sizing workflow. The sizing skill bundles an estimator script that computes the RAM/disk/VRAM figures directly from your rows, dimensions, and index choice.

Get started

Install from the public KX skills marketplace — the same plugin, and the same skills, install into either client.

# Type these inside a Claude Code session
/plugin marketplace add https://github.com/KxSystems/kx-skills.git
/plugin install kdbai-knowledge@kx-skills

Restart Claude Code, or start a new session, so it picks up the plugin's skills.

# Run from your shell
codex plugin marketplace add https://github.com/KxSystems/kx-skills.git
codex plugin add kdbai-knowledge@kx-skills

Restart Codex, or start a new session, so it picks up the plugin's skills.

Refer to the kx-skills repository for the full list of KX plugins, update instructions, and release notes.

Ask a question

Once the plugin is installed, ask in natural language from the editor or terminal. For example:

  • Which index should I use for 5 million 1,536-dimension vectors with streaming inserts?
  • How much RAM and disk will 10 million 768-dimension vectors need on qHNSW?
  • Write the schema for a table carrying both dense and sparse vectors for hybrid search.

Because the answers come from the bundled skills, the index names, parameters, and type names returned match this documentation.

Next steps

  • Size a Deployment covers the workflow the sizing skill follows, and the figures it reports.
  • Use Indexes gives the index comparison behind its schema guidance.
  • Manage Tables shows schema and index definition in the Python, REST, and q APIs.