From Hackathon to Investor Deck in 12 Questions

Where this fits

You just finished a hackathon build and a judge asks "where's your deck?" — or you're prepping for demo day with a working prototype and zero slides. The pitch-deck skill turns that moment into a structured workflow: it interviews you, reads your existing context files, picks an audience-appropriate narrative, and generates slides with speaking notes.

Why agents benefit

• Runs a 12-question interview one question at a time, so the agent extracts real product details instead of generating a generic template.
• Detects the audience (investors, grants, hackathon judges) and adjusts narrative tone and slide order accordingly.
• Reads .claude/context/idea.md and .claude/context/build.md to pre-fill problem, wedge, traction, and stack — no re-asking.
• Self-scores the output and preps objection responses, turning slide generation into pitch coaching.
• Produces speaking notes per slide, so the agent output is immediately usable on stage.

Practical scenarios

Hackathon finals, 15 minutes away

You built a Solana program over the weekend and the final was just announced. Run the skill, answer 12 quick questions, and get a 10-slide deck with demo-ready talking points. The agent prompts you for a live demo link and builds the slide order around it.

Grant application for the Solana Foundation

Grant reviewers read 50 decks a week. The skill flips the narrative to problem → wedge → traction → team, skipping the "what is blockchain" intro. It pulls traction numbers from your context files so you don't copy-paste.

Accelerator application with a co-founder

You and your co-founder are on a call; one of you has context files, the other has the numbers. The skill merges both — it reads the context, then asks only for what's missing, so you finish in under 20 minutes.

Add it to your agent workflow

Install with:

npx skills add solanabr/solana-ai-kit --skill pitch-deck

Then trigger it in conversation:

User: "We need a demo day deck for our DePIN project."
Agent: Reads .claude/context/idea.md, starts the 12-question interview.

The agent walks through the interview, then outputs a complete markdown deck with speaker notes, a self-score, and an objections section.

Compare before adopting

Check three things before you commit. Quality signals: the skill is adapted from sendaifun/solana-new (MIT licensed) with telemetry removed — good community hygiene. Maintenance: 97 stars on the solana-ai-kit repo, but check the last commit date; crypto moves fast. Alternatives: you could prompt a generic LLM to write slides, but you'd lose the structured interview, context-file handoff, and audience detection. For Solana-specific projects, this is the best workflow fit.

Why it is worth tracking

The repo has 97 stars and sits in an active Solana AI kit ecosystem — a signal that the maintainers are shipping. The skill's design pattern (interview → audience detection → narrative → self-score) is worth stealing even if you build your own. Evaluate it when you're about to pitch, and track it if you're building agent tools for crypto builders. See the full listing at OpenAgentSkill.