Launching today

Amaroad
Build AI slide decks without leaving your terminal
17 followers
Build AI slide decks without leaving your terminal
17 followers
Amaroad is an open-source AI-first slide authoring environment for developers. Ask Claude Code, Codex, Gemini CLI, or Cursor to create and refine decks from your terminal, preview slides live, edit multiple slides in parallel, and export to PDF/PPTX or share a live URL.











Hey there! π
I'm Masato, founder of CORe Inc.
I spend my whole day building with AI coding agents. Claude Code, Codex, Gemini CLI, Cursor. But the moment I needed a deck for a sprint review or a pitch, my flow broke. Leave the terminal, open PowerPoint, drag boxes around by hand. That context switch killed my momentum every time. So I asked: what if the same agent that writes my code could write my slides too?
That's Amaroad. You stay in the terminal, say "create 8 slides for tomorrow's review from this project," and the deck shows up in your browser with live reload. Slides are MDX files, so the agent revises them exactly like code. After the first draft, you run one agent per slide and push a dozen revisions in parallel instead of fixing them one by one.
What I really want to show off is how much is built in. Amaroad ships with 18 agent skills, so you do almost everything just by talking to the agent:
Generate in-slide images with Nanobanana (Gemini) or GPT/Codex imagegen, then edit them later with the same prompt-based flow
Draw theme-aware SVG diagrams: flowcharts, architecture maps, comparisons
Turn an Excel sheet into a deck, row by row
Make graphic-recording style illustrations from your slide content
Export the whole deck to an animated video with Remotion
Translate slides between Japanese and English while keeping the MDX intact
Auto-fix overflow, run a preflight lint, polish speaker notes, fact-check numbers, normalize theme colors
When the deck is done, share a live URL via Cloudflare Tunnel, jump into presenter mode, or export to PDF/PPTX.
It started as an internal tool for our team, but it changed how we work so much that we open sourced it under MIT.
Would love your honest feedback!