#python

Fill an existing AcroForm PDF's real form fields from supplied data and flatten the result into a finished, non-editable PDF

Discover, document, and monitor what the agent harness can do — Python libraries, tools, MCP servers, and runtime capabilities — with repeatable, comparable snapshots.

Generate a silent, looping 1920x1080 MP4 for a conference booth, kiosk, or lobby screen — rendered frame-by-frame in Python (Pillow + ffmpeg), with previews before the full render.

Play a deterministic, resumable Colossal Cave adventure with exact game output and built-in diagnostics.

Run Monte Carlo simulations from natural-language risk inputs — triangular, normal, uniform, or log-normal — and return percentiles, a histogram PNG, optional interactive HTML, and a downloadable results spreadsheet.

Creates polished PowerPoint decks from a JSON specification using python-pptx. Designed as a Copilot Studio Skill that runs natively inside the agent's Python container (no Azure Function or custom connector required). Supports 8 layouts plus native charts (bar, column, line, pie, donut with stacked / 100% stacked variants).

Visualize routes and locations on a map. Accepts pre-ordered stops with coordinates from upstream connectors (Azure Maps, Bing Maps, Dataverse, CRM) plus optional road geometry, leg distances, and durations. Produces PNG, interactive HTML (Leaflet/OSM), GeoJSON, KML, deep links, and QR codes. Fully offline Python engine — road geometry rendered directly when provided, or OSRM used browser-side as a schematic fallback.