A Datama Skill teaches your AI assistant how to deliver interactive Datama visualizations as standalone HTML, without sending your data to Datama servers.


1. What is an AI Skill?

An AI Skill (also called Agent Skill) is a portable package of instructions, reference files, and assets that a compatible AI assistant can read and follow.

Instead of explaining Datama configuration rules in every conversation, you install the skill once. The assistant then knows:

  • how to interpret your data and your question,
  • which Datama solution to use,
  • how to build a valid configuration,
  • how to deliver a ready-to-open HTML page.

The skill package is a folder (or zip) that typically contains:

File Role
SKILL.md Main instructions for the AI assistant (workflow, rules, routing)
references/ Datama configuration specifications (Compare, Explore, market equation)
assets/ HTML output template used to render the visualization
scripts/ Optional validation tools for the generated output
README.md Human-readable overview of the package


2. What does the Datama skill do?

The datama skill teaches a compatible assistant to produce a standalone HTML page with an interactive Datama visualization. All rendering happens in the browser, your dataset stays in the page.

The skill supports two solutions. The assistant picks one per delivered page:

Compare: variance decomposition / waterfall

  • Intent: explain why a KPI changed between two sides (periods, segments, groups: before/after, A vs B, X vs Y).
  • Typical asks: “why did revenue drop?”, “Compare Q1 vs Q2”, “decompose the conversion change”, “delta mobile vs desktop”.
  • Output: interactive waterfall chart with variance decomposition.

Explore: breakdown / slicing

  • Intent: slice a metric across one or two dimensions, or inspect a funnel step: without a mandatory two-group comparison.
  • Typical asks: “break down revenue by country”, “conversion by device and channel”, “sessions per campaign”, “explore the funnel at checkout”.
  • Output: interactive breakdown / slicing visualization.

Routing (how the assistant chooses)

  1. Binary comparison (two periods, two segments, “vs”, “before/after”, “why did X change”) → Compare.
  2. Funnel step, single metric over a dimension, breakdown, or distribution without a mandatory two-side comparison → Explore.
  3. If both comparison and breakdown language appear → Compare (priority: compare > explore).
  4. If unclear → Compare.

Beyond the chart

  • Look and feel: palettes, custom colors and style presets can be requested in plain language. Learn more
  • Results as text (licensed skill): the assistant can read back the computed drivers table and the generated comment, on claude.ai through the artifact itself. Learn more
  • Download as HTML from the header of the delivered page. Learn more


3. Security and license

  • Your data stays in the browser. The dataset is inlined in the HTML page; rows are not uploaded to Datama for rendering.
  • Your license is verified when you download the skill with a license included. Generated visualizations run in paid mode.


4. Compatible hosts

The Datama skill works with any Agent Skills–compatible assistant host, for example Claude, ChatGPT, Cursor, Copilot, or other products that support skills from a zip or folder.

  • Web assistant: upload the datama.zip in your account or project skills (the exact menu varies by product).
  • IDE or desktop agent: copy the skill folder into your host’s skills directory (for example .agents/skills/).


5. Next steps