Reports & PDFs

Turn data from across your workflow into a polished Markdown report, then render it as a nicely formatted PDF — and save it. No templating code, no manual formatting.

The pipeline

Sources (incl. a Store node) → Report → PDF → Save

Report node

Wire one or more upstream nodes into the Report node and it gathers their outputs as sources, then writes a report over them. It can also pull from a Vector Store / Knowledge Base node connected to it — the report searches that store as a tool, so it can cite your own documents. The node shows how many sources it gathered (and which) as you build.

Configuration

• Template — the shape of the report:
  • Auto — let the model choose a sensible structure.
  • Executive summary — short, decision-oriented.
  • Data analysis — findings over the data you fed in.
  • Comparison — compare options/items side by side.
  • Custom prompt — write your own instruction ("Write a report that…").
• Model — any model from your registry (e.g. gpt-4o-mini, apple-intelligence-text, cli-claude-code). On-device Apple Intelligence falls back to cloud in headless/webhook runs.
• Output format — Markdown (default) or HTML.
• Length — Short (~300 words), Medium (~700), or Long (~1500).
• Sections — optional comma-separated outline, e.g. Overview, Findings, Recommendations.
• Include raw data appendix — append the source data at the end.

Output: the report as Markdown (or HTML), ready to feed into the PDF node — or into any other node via {{input.markdown}}.

PDF node

Feeds on the Report node's Markdown (or any Markdown/HTML) and produces a formatted PDF.

Configuration

• Engine
  • Auto (recommended) — picks the best renderer available on your setup.
  • High quality — print-quality rendering with exact page layout. Built into the Circuitry desktop app; on the web or iPad it uses a paired Circuitry Server.
  • Built-in — renders right inside Circuitry on any device, no setup needed. Output is a little simpler.
• Page size — A4, Letter, or Legal.
• Orientation — Portrait or Landscape.
• Theme — Default, Minimal (sans-serif), or Professional (serif).
• Filename — optional; defaults to the source name ({{__source.stem}}.pdf).

Output: the rendered PDF (with its page count), ready to save, email, or return from a webhook.

Save the PDF

Add a Save node after the PDF node and pick an output folder. Each run writes the PDF there (using the filename from the PDF node, or a naming rule you set). Saving local files uses your project folder or a paired Circuitry Server.

Build it

1. Wire your sources into a Report node — data nodes, and a Store node if you want it to draw on your knowledge base.
2. Pick a Template, Model, and Length; optionally list Sections.
3. Connect Report → PDF. Choose a Theme, Page size, and Engine.
4. Connect PDF → Save and choose an output folder.
5. Run. You get a formatted PDF in your folder. Add a Then node to do something once it's done (email it, post a link, notify a channel).

Examples

• Weekly KPI report — Sheet + API sources → Report (Executive summary) → PDF (Professional, A4) → Save → Then (email it).
• Knowledge-base brief — Store node + a question → Report (cites your docs) → PDF → Save.
• Research digest — several web/API sources → Report (Data analysis, Long) → PDF (Minimal) → Save.

Next steps

• Knowledge Base — add a Store node as a report source
• Batch File & Image Processing — another file-producing pipeline
• Circuitry Server — high-quality PDF rendering + local saves when using Circuitry on the web or iPad