Hypeplan

Listen up bozo, you got two options:

1: Keep reading your shitty text-based plans.

2: Have your plan delivered to you as an ASCII keynote speech in your terminal.

If you really think there's a choice here, you're ngmi.

What the...?

Hypeplan takes a normal markdown-based plan and hypes it.

Suddenly the plan has its own stage, a keynote speaker, an audience and becomes infinitely more convincing.

I still don't...

Jesus, just look...

... or watch

But why

• Because normal projects already exist.
• I asked "what if?" exactly once.
• There was a lack of adult supervision.
• And it's now way too late for that question buddy.

I'm down, what do I need?

• Bun
• An OpenAI API key
• An audio player: afplay on macOS, or ffplay elsewhere

Install

git clone git@github.com:martinrue/hypeplan.git
cd hypeplan
bun install
bun dist
echo "OPENAI_API_KEY=your-openai-api-key" > .env
cat plan.md | ./hypeplan

Arguments

Hypeplan supports these flags:

• --voice <name> Use a specific TTS voice. Supported built-in defaults are coral, marin, and cedar.

• --style <name> Choose the presentation style prompt. The default is product-keynote.

• --script-model <name> Override the model used to generate the presentation script. Default: gpt-5.4-mini

• --tts-model <name> Override the model used to generate speech audio. Default: gpt-4o-mini-tts

• --tts-speed <number> Set speech speed to any positive number. Default: 1

• --no-cache Regenerate the script and audio instead of using cached assets.

Runtime controls

• Right Arrow Skip to the next segment of the keynote

• Left Arrow Go back to the previous segment

• Esc, Ctrl-C, q Quit

The plan

Hypeplan expects a markdown-based plan to be piped in on stdin or to be present on the clipboard.

To check it's a valid plan, hypeplan looks for:

• a heading
• bullets or numbered items
• plan-oriented language such as summary, implementation, scope, rollout, or test plan

Caching

Generated assets are cached under .cache/.

Use --no-cache if you want to force regeneration.