talk

Orchestrates a six-stage pipeline that builds conference talks from inputs to a built React slide deck.

1|Updated Nov 25, 2021
One-click install
npx skills add https://github.com/rstagi/dotfile --skill talk-rstagi
Or copy as Structured Prompt for Agentโ–ผ
Please help me install this Agent Skill.
Skill: talk
Source: https://github.com/rstagi/dotfile/tree/main/.claude/skills/talk
Command: npx skills add https://github.com/rstagi/dotfile --skill talk-rstagi

SYSTEM DOCUMENTATION & REQUIREMENTS

๐Ÿ’ก This Skill includes references (resource) components.

What problem does it solve? Building a conference talk involves many disconnected steps โ€” shaping the argument, structuring slides, writing speaker notes, styling the deck, building it, and checking consistency โ€” and doing them ad hoc produces generic, AI-sounding results. This Skill gathers the talk's inputs once and drives six sub-skills in order, with anti-slop quality gates between every stage. ## Core Features & Use Cases - End-to-end pipeline orchestration: Runs content brief, slide structure, delivery plan, slide styling, deck build, and refinement as six sequential stages, each producing a versioned artifact in a shared working directory. - Keep-It-Human quality gates: Applies Gates Aโ€“E from the bundled references between stages, refusing to advance when 3+ flags (4+ for visual checks) accumulate, and verifies every statistic against a fact-checked bibliography. - Resume and standalone modes: Detects existing artifacts in the working directory and resumes from the furthest completed stage, or runs any single stage (content, slides, delivery, style, build, refine) on its own. - Use Case: You are speaking at a conference in three weeks. Provide your topic, talk type, time slot, audience, venue, and point of view, and the pipeline produces a content brief, assertion-titled slide structure, cue cards, a styled custom React deck, and a ship-readiness report. ## Quick Start Ask the assistant to help you build a conference talk from your idea, providing the topic, talk type, time limit, audience, venue, and your point of view.

Frequently Asked Questions about talk

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a conference talk from scratch with AI?โ–ผ

Provide six inputs โ€” topic, talk type, time, audience, venue, and point of view โ€” and the pipeline runs six stages in order: content brief, slide structure, delivery plan, slide styling, deck build, and refinement. Each stage writes an artifact to a working directory like ~/talks/<slug>/.

Can I run only one stage of the talk pipeline?โ–ผ

Yes. Each stage is a standalone sub-skill: talk-content for the argument, talk-slides-content for slide structure, talk-delivery for cue cards, talk-slides-style for the look, talk-slides-build for the deck, and talk-refine for final consistency checks.

What format is the built slide deck?โ–ผ

The default output is a custom React deck built with Vite and Framer Motion, with per-slide speaker notes and keyboard navigation. A reveal.js Markdown deck with PDF export is available as a documented fallback when a vector PDF is required.

What happens if my talk working directory already has files?โ–ผ

The pipeline detects existing artifacts such as 01-content-brief.md or the deck directory, lists what exists, infers the furthest completed stage, and offers to resume from the next stage rather than overwriting your work.

How does the pipeline prevent generic AI-sounding slides?โ–ผ

It runs Keep-It-Human gates between stages checking specificity, voice, slide visuals, and delivery against a shared rubric. A stage that trips 3 or more flags (4 for visual checks) must be rewritten before the pipeline advances.

When should I not use the full talk pipeline?โ–ผ

Skip it when you only need one piece, such as just the argument or just the styling โ€” invoke that sub-skill directly instead. Also do not start without a topic and inputs, since the pipeline asks for them up front rather than inventing a talk.