All flashcards — dashboard from data

A named, reusable agent capability defined in plain language. Turns a multi-step workflow (read data, compute, render dashboard) into a one-line invocation any authorized user can call.

By processing large lists with sandboxed code instead of stuffing rows into the prompt, the agent handles SharePoint datasets of any size. The model only holds the code and the summary, never the raw data.

The worker-agent's signature move: write actual code, run it in a sandbox, use the computed results. Numbers in the dashboard come from real execution, not LLM arithmetic — which is what makes them trustworthy.

Asking the agent to reshape an existing artifact (e.g. HTML dashboard) into a different format (PowerPoint, PDF). Reads the source, emits the target — no re-running the analysis pipeline.

The demo scenario: messy SharePoint lists of candidates and open roles, a one-sentence invocation, and a polished interactive HTML dashboard back — assembled by an AI Skill the user defined earlier.

The Skill's output: a self-contained HTML file with embedded JavaScript dropped into SharePoint. Interactive, shareable, archivable, re-openable — a durable artifact, not a chat reply that vanishes.

Layering Skills by output format: one Skill produces analytical HTML, another remixes it into an executive PPT, another a one-pager. Skills compose into pipelines; each stays small and tunable.

The actual reason to package work as a Skill. Locks the output shape so quarter-over-quarter comparisons are visually direct. The data changes every run; the structure does not.

A flow visualization showing how candidates move through hiring stages. Renders properly in HTML thanks to JavaScript; falls back to a bar chart in PPT because the format can't host the interactive view.

The author's spec inside a Skill: data sources, metrics that matter, output shape (header, metric bar, charts), and explicit stages or thresholds. Opinionated about structure, flexible about data.