presentation-creation
Create polished, audience-tailored presentations from outlines or documents, with support for Marp, reveal.js, Google Slides, and PPTX output formats. Use when the user requests presentation creation or provides relevant inputs for this workflow.
git clone --depth 1 https://github.com/seb1n/awesome-ai-agent-skills /tmp/presentation-creation && cp -r /tmp/presentation-creation/communication/presentation-creation ~/.claude/skills/presentation-creationSKILL.md
# Presentation Creation This skill enables an AI agent to transform raw content — outlines, documents, meeting notes, or data — into professional presentation slide decks. The agent applies slide design principles such as one idea per slide, visual hierarchy, and minimal text to produce decks that are clear, engaging, and presentation-ready. It supports multiple output formats including Marp Markdown, reveal.js HTML, Google Slides, and PPTX. ## Workflow 1. **Gather content and define the audience.** Collect the source material — an outline, a technical document, meeting notes, or a set of talking points. Identify the target audience (executives, engineers, customers, investors) and the presentation goal (inform, persuade, train, pitch). The audience determines vocabulary depth, level of detail, and visual style. An investor pitch uses bold claims and metrics; a technical deep-dive uses architecture diagrams and code snippets. 2. **Structure the narrative arc.** Organize the content into a logical story: opening hook, problem statement, solution or key message, supporting evidence, and closing call to action. Map this arc onto a slide-by-slide outline. A typical 15-minute presentation uses 10-15 slides. Each slide should convey a single idea — if a slide needs two headings, it should be two slides. Add transition beats between sections to maintain narrative flow. 3. **Write slide content with minimal text.** For each slide, write a concise headline (under 8 words) that states the takeaway, not the topic. Use "Revenue grew 40% YoY" instead of "Revenue Overview." Keep body text to 3-5 bullet points or a single short paragraph. Where possible, replace text with visuals — a chart, diagram, screenshot, or image that communicates the point faster. Write speaker notes for each slide with the full talking points the presenter should deliver verbally. 4. **Suggest visual elements and layout.** For each slide, recommend the appropriate visual treatment: bar chart for comparisons, line chart for trends, flow diagram for processes, screenshot for product demos, icon grid for feature lists, or a full-bleed image for emotional impact. Specify layout as title-only, title-and-body, two-column, image-left, or full-image. Suggest a consistent color palette and font pairing (e.g., Inter for headings, Source Sans for body). 5. **Generate the slide deck in the target format.** Produce the complete deck in the requested format. For Marp, output a single Markdown file with `---` slide separators and Marp directives. For reveal.js, output an HTML file with section tags. For PPTX, generate a python-pptx script or direct file. For Google Slides, output a structured JSON payload compatible with the Slides API. Include speaker notes, slide numbers, and any embedded media references. 6. **Review and refine.** Check the deck against presentation best practices: no slide exceeds 40 words of body text, every slide has a clear takeaway headline, visual variety prevents monotony (not all bullet-point slides), the opening slide grabs attention, and the closing slide has a clear next step. Adjust pacing — if the deck is too long, identify slides that can be merged or moved to an appendix. ## Usage Provide the agent with the source content (or a description of the topic), the target audience, desired slide count, and output format. The agent will return a complete slide deck with speaker notes. **Prompt format:** ~~~ Create a presentation about [topic]. Source material: [outline, document path, or key points] Audience: [who will see this] Slide count: [target number of slides] Output format: [marp / revealjs / pptx / google-slides] ~~~ ## Examples ### Example 1: Product Launch Slide Deck (Marp Format) **Input:** ~~~ Create a presentation for the launch of "Nimbus," a cloud cost optimization tool. Audience: Engineering managers and VPs at mid-size SaaS companies. Slide count: 7 Output format: marp ~~~ **Output:** ~~~markdown --- marp: true theme: default paginate: true --- # Nimbus ## Cut your cloud bill by 35% — automatically  <!-- Speaker notes: Welcome everyone. Today I'm excited to introduce Nimbus, our new cloud cost optimization platform. By the end of this presentation, you'll see how Nimbus can save your team six figures annually without any code changes. --> --- # The Problem: Cloud Spend Is Out of Control - 73% of engineering leaders say cloud costs exceeded budget last quarter - Idle resources, oversized instances, and forgotten dev environments - Manual cost reviews happen quarterly at best — waste compounds daily <!-- Speaker notes: Let's start with the pain. If you're like most teams we talk to, your cloud bill is a source of constant surprise. The problem isn't that you're using too much cloud — it's that nobody can see where the waste is in real time. --> --- # How Nimbus Works 1. **Connect** — Link your AWS, GCP, or Azure accounts in under 5 minutes 2. **Analyze** — Nimbus scans utilization, traffic patterns, and billing data 3. **Optimize** — Get prioritized recommendations ranked by savings impact 4. **Automate** — Enable auto-scaling rules and scheduled shutdowns <!-- Speaker notes: Nimbus works in four simple steps. The onboarding is read-only and non-invasive. No agents to install, no IAM changes beyond read access. --> --- # Results: What Our Beta Customers Achieved | Company | Monthly Savings | Time to ROI | |---------------|-----------------|-------------| | Acme Corp | $42,000 (31%) | 2 weeks | | DataFlow Inc | $67,000 (38%) | 3 days | | Bright SaaS | $28,000 (29%) | 1 week | Average: **35% cost reduction** within the first month. <!-- Speaker notes: These aren't projections — these are real numbers from our beta cohort. DataFlow saw results in 3 days because they had a large cluster of dev instances running 24/7. --> --- # Built for Engineering Teams, Not Finance - No code changes required — works
Design reproducible evaluations for AI agents with representative task sets, explicit rubrics, appropriate graders, baselines, regression gates, and failure analysis. Use when defining agent quality, comparing prompts or models, validating a release, measuring tool-use reliability, investigating regressions, or deciding whether an agent is ready for production.
Design privacy-aware observability for AI agents using traces, spans, structured events, metrics, cost attribution, dashboards, alerts, and investigation workflows. Use when instrumenting an agent, debugging intermittent tool or model failures, defining service-level objectives, analyzing latency or spend, auditing agent decisions, or preparing production monitoring.
Design and verify auditable human oversight, approval gates, escalation paths, and safe state transitions for AI agent workflows. Use when deciding which agent actions require review, adding approve/reject or dual-control flows, preventing unauthorized autonomous effects, creating decision records, reducing rubber-stamping, or recovering safely from rejected, expired, or failed actions.
Design, implement, harden, and verify Model Context Protocol (MCP) servers with precise tool contracts, least-privilege authorization, safe transports, structured errors, and interoperability tests. Use when creating a new MCP server, exposing an API or data source through MCP, reviewing an MCP server design, adding or revising MCP tools, or preparing an MCP server for production.
Design and operate bounded multi-agent workflows with task decomposition, dependency graphs, ownership, handoff contracts, shared-state controls, approvals, recovery, and synthesis. Use when a task contains genuinely independent workstreams, specialized roles, parallel research or implementation, reviewer-worker loops, or coordination problems that one agent should not execute sequentially.
Design and validate model-facing tool definitions with clear names, action-oriented descriptions, bounded JSON Schema parameters, explicit side effects, safe defaults, idempotency, errors, and realistic tests. Use when creating function-calling tools, MCP tools, agent actions, structured tool inputs, or when a model selects the wrong tool, invents arguments, or causes unsafe side effects.
Plan, execute, document, and retest authorized security assessments of AI agents and multi-agent workflows using safe adversarial cases, synthetic identities, canaries, and evidence-based findings. Use when defining red-team rules of engagement, assessing prompt injection or excessive agency, testing tool and identity boundaries, evaluating memory or cross-agent attacks, scoring a campaign, or verifying remediation in an approved environment.
Threat-model and harden AI agents, RAG systems, assistants, and tool-using workflows against direct, indirect, stored, cross-agent, and multimodal prompt injection. Use when reviewing an agent architecture, isolating untrusted content, constraining tools and egress, protecting secrets, adding injection-focused tests, investigating a suspected injection incident, or documenting residual prompt-injection risk.