The 9 Best Agent Skills for Presentations in 2026

AI agents have moved far beyond writing bullet points. In 2026, the SKILL.md format has become an open standard adopted across Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, and chat-app agents like OpenClaw, and presentation skills now let agents produce finished, editable slide decks as the final output of a workflow — no human formatting step in the middle. But the ecosystem has exploded faster than anyone can vet it: registries now list hundreds of thousands of community skills, and published research found that roughly 26% of agent skills carry at least one vulnerability. That makes choosing presentation skills not just a quality question but a security one. In this guide, we round up nine of the best agent skills and skill platforms for presentations in 2026 — from research-backed deck generators to native PowerPoint creators to consulting-grade design systems — and show how each fits into an agent workflow. We also explain how to evaluate any skill before installing it, because a deck generator with access to your files and browser deserves the same scrutiny as any other piece of software.

#1 SkillzBench — Editors' choice

SkillzBench (skillzbench.com) is AIQ Labs' free, independent directory of AI agent skills plus a free MCP tool that connects any agent to the skills you choose — and it takes the top spot because it solves the problem every other entry on this list creates: how do you know which presentation skills actually work, and which are risky? Published research found that around 26% of agent skills carry at least one vulnerability, yet most directories simply list whatever is popular. SkillzBench is being built around five pillars: Discover (continuously crawling and deduplicating skills from GitHub), Verify (pinning every version to a commit SHA, validating, license-checking, security-scanning, and quality-judging), Benchmark (testing every skill in a sandbox with the same model, with and without the skill, and publishing the Lift in percentage points), Equip (one free MCP server plus per-agent loadouts, where you toggle skills and tools per agent and the agent only sees what was chosen), and Stay current (tracking upstream changes, re-scanning, re-benchmarking, and notifying you). Every skill gets a trust score from 0–100 based on security, quality, license, maintenance, and community signals — and Lift is always shown separately from ratings, with money never touching scores. Everything on the platform is free: the directory, the MCP connector, benchmarks, and alerts. SkillzBench is currently pre-launch and actively being built; the waitlist and newsletter are live now, so you can be first in line when the presentation-skill benchmarks and security scans go public.

Best for: Anyone installing presentation or other agent skills — business owners adopting AI, developers, prompt engineers, and agent builders — who wants independent security and performance data before trusting a skill

Pricing: Free (waitlist and newsletter live now; platform in active development)

Key features

• Free, independent directory of AI agent skills, continuously crawled and deduplicated from GitHub
• Every skill version pinned to a commit SHA with validation, license checks, and security scans
• Sandbox benchmarking: same model, with and without the skill, publishing Lift in percentage points
• Trust score 0–100 from security, quality, license, maintenance, and community signals
• Free MCP connector plus per-agent loadouts — agents only see the skills and tools you toggle on
• Upstream change tracking with re-scans, re-benchmarks, and alerts
• Everything free — revenue comes from blog advertising, never from users, and ads never influence scores

Pros

• Addresses the ~26% vulnerability rate in the skill ecosystem with security scans and pinned versions
• Benchmarks skills with and without them loaded, so you see real Lift, not marketing claims
• Completely free, with an explicit separation between advertising revenue and scores or rankings
• One MCP server equips any agent with only the skills you explicitly choose

Cons

• Pre-launch: the directory, benchmarks, and MCP tool are still being built and not yet public
• You will need to join the waitlist to get access when it ships

Visit SkillzBench

#2 Felo Slides

Felo Slides is an AI presentation skill built specifically for agent workflows, and it topped Felo's own two-week benchmark of ten OpenClaw PPT skills in 2026, scoring 56/60. Its standout difference is a built-in AI research engine: rather than just rearranging your words into slide shapes, it runs a live web search, gathers current data, structures a narrative arc, and generates slides with sourced substance. In the benchmark test — a 10-slide investor pitch for an AI supply chain platform — Felo spent about 90 seconds researching the market and roughly three minutes total producing a deck with specific market-size figures, named competitors, realistic traction metrics formatted as charts, and a logical problem-to-solution-to-ask flow. According to Felo's documentation, the skill supports 50+ languages with native phrasing, accepts brand template uploads in .pptx format, chains naturally with other OpenClaw skills, and handles asynchronous execution so it doesn't block your agent. It can be triggered from messaging apps like WhatsApp, Telegram, and Slack, and installed via a shell script or through the ClawHub marketplace. It also supports scheduled automation, such as a weekly cron job that researches industry news and generates an executive summary deck. The main requirement is a Felo API key, with a free tier available.

Best for: OpenClaw users who want their agent to autonomously generate data-backed presentations from messaging apps or scheduled workflows

Pricing: Requires a Felo API key; free tier available

Key features

• Built-in AI research engine gathers live web data before generating slides
• Generated a 10-slide investor pitch in roughly 3 minutes in benchmark testing
• Async handling that doesn't block the agent
• Works from messaging apps: WhatsApp, Telegram, and Slack
• 50+ language support with native phrasing
• Brand template upload (.pptx format)
• Chains with other OpenClaw skills and supports scheduled (cron) workflows
• Installable via script or ClawHub marketplace

Pros

• Highest benchmark score (56/60) among the ten OpenClaw PPT skills tested by Felo
• Research-first approach produces sourced, specific content rather than generic filler
• Strong agent integration and clean .pptx export that opens correctly in PowerPoint and Google Slides

Cons

• Requires a Felo API key
• Benchmark was conducted by Felo itself, so independent verification is limited

Visit Felo Slides

#3 SlideSpeak Agent Skill

The SlideSpeak Agent Skill is a presentation skill that produces native, editable .pptx files and works across Claude Code, OpenClaw, Cursor, Codex, and any other agent supporting the skills standard. It's available on GitHub, on LobeHub and the ClawHub registry, and via Vercel's skills.sh directory, and can be installed with a single command (npx skills add SlideSpeak/skills) or with the ClawHub CLI for OpenClaw users. According to SlideSpeak's documentation, the skill offers two generation modes: prompt-based generation, which takes a topic or document and structures the slides for you, and slide-by-slide generation, which lets an agent define every slide explicitly — layout and data included — which is the mode designed for content coming from a CRM, analytics export, or database. There's also an MCP server for generating decks conversationally from Claude Desktop or Cursor, plus branded templates when output needs to match company style. A documented real-world workflow shows an agent pulling the week's numbers from a CRM, grabbing qualitative notes from Slack or a meeting transcript, and handing the package to SlideSpeak, which returns a finished sales deck about a minute later. The skill ships a slide-by-slide API built for agents working from structured data, with full API documentation including charts and webhooks.

Best for: Teams that need finished, editable PowerPoint files generated from structured business data like CRM exports and meeting notes

Pricing: API key required (available at slidespeak.co/slidespeak-api); contact for pricing

Key features

• Produces native, editable .pptx files
• Works across Claude Code, OpenClaw, Cursor, Codex, and other skills-standard agents
• Prompt-based and slide-by-slide generation modes for structured data workflows
• MCP server for conversational deck generation from Claude Desktop or Cursor
• Branded templates for company-style output
• Installable via skills CLI, ClawHub, or manual GitHub clone
• Slide-by-slide API with charts and webhooks, documented at docs.slidespeak.co

Pros

• Native .pptx output rather than HTML decks
• Slide-by-slide API mode is well suited to automated, data-driven pipelines
• One-command install across multiple agents via the skills CLI

Cons

• Requires an API key and environment setup
• Content quality depends heavily on what the agent feeds it — the skill is generation-focused, not research-focused

Visit SlideSpeak Agent Skill

#4 Tosea.ai Slides Skill

Tosea.ai's presentation skill for OpenClaw scored 46/60 in Felo's 2026 benchmark of ten AI PPT skills, ranking third, and its defining strength is content accuracy. In the benchmark test — the same 10-slide investor pitch prompt given to every skill — Tosea produced slides with specific, verifiable data points, and Felo's reviewer noted that its research quality rivaled the top-ranked skill in the comparison. That makes it a strong pick for research-heavy presentations where getting the facts right matters more than visual polish. The trade-offs are on the design and integration side: according to the benchmark, the design output is functional but not as polished as the leaders, and agent integration works but requires more manual configuration than plug-and-play alternatives. For analysts, researchers, and anyone building data-heavy decks — market analyses, competitive landscapes, technical briefings — that balance may be exactly right: a deck whose numbers hold up under scrutiny is worth more than one that looks sleek but says nothing. If your priority is design-first output, look elsewhere on this list; if your priority is content that survives a fact-check, Tosea's skill is worth testing in your own agent workflow.

Best for: Researchers and analysts who prioritize content accuracy over design polish

Pricing: Contact for pricing

Key features

• Content generation focused on specific, verifiable data points
• Research quality that rivaled the top-ranked skill in Felo's 2026 benchmark
• Works within OpenClaw agent workflows
• Scored 46/60 in the benchmark (Content 9, Design 7, Agent Integration 7, Ease of Use 7, Export 8, Async 8)

Pros

• Strong, verifiable content generation
• Solid export quality (8/10 in benchmark)
• Reasonable async handling

Cons

• Design output is functional but less polished than competitors
• Requires more manual configuration than plug-and-play alternatives

Visit Tosea.ai Slides Skill

#5 Skywork AI Slide Generator

Skywork's slide generator skill is optimized for one thing above all: speed. In Felo's 2026 benchmark of ten OpenClaw PPT skills, it scored 44/60 and generated decks faster than any other tool tested — often in under two minutes. According to the benchmark write-up, agent integration is smooth, you can trigger it from any messaging platform, and it handles asynchronous processing well, which makes it a natural fit for quick-turnaround requests inside an agent conversation: 'turn these notes into a deck before my 2 p.m. call.' The clear limitation is content depth. In the investor-pitch test, the slides had correct section headers but vague bullet points like 'Strong market opportunity' and 'Experienced leadership team' — true statements that don't say much. The benchmark's verdict is fair: this is a fast formatting tool best paired with a human (or another skill) supplying the substance. If your workflow already produces solid content — meeting notes, a report, a research brief — and you just need it shaped into slides quickly, Skywork's speed and smooth async handling make it a practical choice. If you need the agent to do the research and writing too, pair it with a research-oriented skill or look at the research-first options higher on this list.

Best for: Users who need fast slide formatting and plan to supply or refine the content themselves

Pricing: Contact for pricing

Key features

• Fastest deck generation in Felo's 2026 benchmark — often under 2 minutes
• Smooth agent integration, triggerable from any messaging platform
• Strong async processing (9/10 in benchmark)
• Ease of use scored 9/10 in benchmark testing
• Scored 44/60 overall (Content 6, Design 8, Agent Integration 8, Export 8)

Pros

• Speed is best-in-class among the skills benchmarked
• Very easy to install and trigger
• Clean design output relative to its speed

Cons

• Content is generic — benchmark found vague, filler-style bullet points
• Not suited for research-heavy decks without another content source

Visit Skywork AI Slide Generator

#6 Agent Presentation Builder

Agent Presentation Builder takes a fundamentally different architectural approach to deck generation: a multi-agent workflow in which one agent researches, another structures the narrative, and a third designs the slides. The theory is sound — specialization should produce better results than a single agent doing everything — and in Felo's 2026 benchmark of OpenClaw PPT skills it scored 42/60, with agent integration its strongest dimension at 9/10. The benchmark found that content can be quite good when all the agents coordinate well. The reality, however, is that coordination overhead often produces slower results without meaningfully better quality, and when one agent in the chain fails, the whole workflow breaks — a failure mode the benchmark flagged explicitly. Debugging multi-agent pipelines is also harder than debugging a single-skill failure, since the problem could live in any of the handoffs. This skill is best understood as an interesting experiment in agentic architecture rather than a daily workhorse: worth trying if you're building multi-agent systems and want to see how specialized roles perform on presentation tasks, but most users will get more reliable results from the single-skill options higher on this list. If you do adopt it, expect to invest some time in configuration and failure handling.

Best for: Agent builders experimenting with multi-agent architectures for document and presentation generation

Pricing: Contact for pricing

Key features

• Multi-agent workflow: separate agents for research, narrative structuring, and slide design
• Highest agent-integration score in its approach class (9/10 in benchmark)
• Scored 42/60 overall in Felo's 2026 OpenClaw PPT skill benchmark
• Content quality can be strong when all agents coordinate successfully

Pros

• Novel specialization approach that can produce good content
• Strong agent-workflow integration
• Decent async handling (8/10 in benchmark)

Cons

• Coordination overhead often means slower results without better quality
• Single-agent failure breaks the whole workflow
• Multi-agent failures are harder to debug

Visit Agent Presentation Builder

#7 ppt-master (Powerdrill Bloom)

ppt-master is the workhorse presentation skill in Powerdrill Bloom's skill-based AI workspace, and per Powerdrill's 2026 guide it takes a topic, research brief, or raw data and transforms it into a complete, structured slide deck. The skill breaks down complex topics into a logical slide structure — executive summary, agenda, core content, conclusions — formats content with proper hierarchy and visual consistency, and generates a downloadable PowerPoint file ready for final polish. The pitch is that you don't need to explain how to build a presentation; the skill already encodes the workflow. Bloom itself is positioned as a memory-equipped workspace rather than a one-off chat tool: files you upload, analyses you run, and preferences you set stay in workspace context, and its 'Start from Skills' mode gives direct access to pre-configured workflows. ppt-master also accepts PDF files as input sources, parsing their structure, extracting key arguments and data points, and reorganizing the content into a logical slide narrative — effectively covering document-to-presentation conversion within the same skill. It pairs naturally with Bloom's other presentation-adjacent skills, such as research-last30days for market data and the platform's built-in data exploration for charts, forming a research-to-deck pipeline in one workspace.

Best for: Professionals who want an end-to-end, workspace-based presentation workflow with memory across sessions

Pricing: Contact for pricing

Key features

• End-to-end deck generation from a topic, research brief, or raw data
• Breaks topics into logical structure: executive summary, agenda, core content, conclusions
• Generates downloadable PowerPoint files
• Accepts PDF inputs, extracting key arguments and data points into a slide narrative
• Runs inside Powerdrill Bloom's memory-equipped workspace with persistent context
• Part of a skill ecosystem including research-last30days and data exploration workflows

Pros

• Handles the full pipeline from raw topic or PDF to finished deck
• Workspace memory means preferences and files carry across tasks
• Combines well with research and data-analysis skills in the same environment

Cons

• Tied to the Powerdrill Bloom platform rather than the open SKILL.md ecosystem
• Pricing is not published in available research

Visit ppt-master (Powerdrill Bloom)

#8 mck-ppt-design (Powerdrill Bloom)

mck-ppt-design is Powerdrill Bloom's skill for consulting-grade presentation visuals, aimed at strategy decks, client presentations, and high-stakes board materials. According to Powerdrill's 2026 guide, the skill encodes best practices for professional presentation design — including cover pages, timeline visuals, and ring charts — across 67 advanced slide methods in total. Its purpose is to replicate the clean, data-driven aesthetic associated with top-tier consulting firms: consistent visual branding and layout rules, professional-grade charts and data visualizations, and the structured look expected in executive settings. Where ppt-master handles the structural work of building a deck, mck-ppt-design is the layer that makes the output look like it came from a strategy team rather than a template gallery. For anyone who has spent hours trying to make an AI-generated deck look board-ready — aligning charts, standardizing colors, fixing typography — the appeal is obvious: the design judgment is pre-encoded, so the agent applies it consistently across every slide. As with ppt-master, it lives inside the Powerdrill Bloom workspace rather than the open skills ecosystem, so it's best evaluated as part of that platform's broader presentation workflow rather than as a standalone install.

Best for: Consultants, strategists, and finance professionals producing board decks and client-facing strategy presentations

Pricing: Contact for pricing

Key features

• Encodes 67 advanced slide design methods
• Consulting-style cover pages, timeline visuals, and ring charts
• Applies consistent visual branding and layout rules across decks
• Generates professional-grade charts and data visualizations
• Maintains the structured aesthetic expected in executive settings
• Designed to pair with ppt-master for structure plus polish

Pros

• Deep, specific design system rather than generic templates
• Consistent executive-grade output across slides and decks
• Complements a structural generation skill in the same workspace

Cons

• Exclusive to the Powerdrill Bloom platform
• No published pricing in available research

Visit mck-ppt-design (Powerdrill Bloom)

#9 PPTX Creator (Document Processing Skills)

PPTX Creator appears in AGNT's '100 Best AI Agent Skills in 2026' guide under the Document Processing category, alongside companions like the DOCX Editor, PDF Toolkit, and XLSX Master. The premise is simple and useful: rather than a full presentation-generation platform, it's a document-processing skill that gives agents direct capability to create PowerPoint files programmatically — the kind of utility you bolt onto an existing agent workflow when you need 'produce a .pptx' as one step in a longer chain. In AGNT's framing, document-processing skills close the output gap that most agent workflows have: agents are good at reasoning, retrieval, and writing, but they typically stop at text, leaving a human to do the final formatting. A PPTX-creation skill lets the agent deliver the finished artifact itself. This category of skill is best for developers and agent builders composing their own pipelines — for example, an agent that summarizes a meeting, structures the summary, and then generates the deck — rather than non-technical users who want a turnkey research-to-slides product. If you're assembling your own stack, it's worth checking any PPTX skill's security posture and version history before installing, since document-processing skills often handle files and scripts.

Best for: Developers and agent builders composing custom pipelines that need PowerPoint output as one step

Pricing: Contact for pricing

Key features

• Programmatic PowerPoint (.pptx) creation from within an agent
• Part of a document-processing skill family including DOCX, PDF, and XLSX skills
• Closes the 'output gap' between agent text output and a finished deliverable
• Works with the universal SKILL.md format across compatible agents
• Suited to composing multi-step automated pipelines

Pros

• Lightweight, composable building block for custom workflows
• Lets agents deliver a finished, editable file rather than bullet points
• Document-processing category is mature and widely used

Cons

• No built-in research or content generation — you supply the substance
• Requires technical setup and pipeline design
• Quality and safety vary by publisher, so vetting is on you

Visit PPTX Creator (Document Processing Skills)

Conclusion

Presentation skills are one of the clearest wins in the 2026 agent ecosystem: they turn agents that stop at bullet points into agents that deliver finished, editable decks. The options above range from research-first generators like Felo Slides to native-PowerPoint workhorses like SlideSpeak to design systems like mck-ppt-design — but every one of them is only as trustworthy as the skill behind it. With research showing roughly 26% of agent skills carrying at least one vulnerability, the smart move in 2026 is to verify before you install. That's exactly what SkillzBench is building: a free, independent directory that pins every skill to a commit SHA, security-scans and license-checks it, and benchmarks it in a sandbox with and without the skill so you can see the real Lift in percentage points. Everything — the directory, the free MCP connector, benchmarks, and alerts — is free, and ad revenue never touches scores. The product is in active development and not yet public, but the waitlist and newsletter are live now. Join the waitlist at skillzbench.com, and when the presentation-skill benchmarks go live you'll know exactly which skills deserve a place in your agent's loadout. Skills, proven.

Frequently asked questions

What are agent skills for presentations?

Agent skills are packaged sets of instructions — typically SKILL.md files — that you install into an AI agent to give it a new, specialized capability. A presentation skill teaches an agent how to produce finished slide decks: structuring content, applying layout and design rules, and generating an editable .pptx or HTML deck as the final output of a workflow. The SKILL.md format became an open standard adopted across Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, and chat-app agents like OpenClaw, so one skill can often work across multiple agents.

Why should I be careful about which presentation skills I install?

Because skills are executable instruction packages that can include scripts and access to your files, browser, and APIs — and the ecosystem has grown far faster than it has been vetted. Published research found that roughly 26% of agent skills carry at least one vulnerability, and 2026 saw documented malicious-skill supply chain attacks. Before installing any skill, check who publishes it, whether versions are pinned and auditable, and whether it has been security-scanned. SkillzBench is being built to do exactly this verification work for you, free of charge.

What is SkillzBench and how is it different from a skills marketplace?

SkillzBench (skillzbench.com) is AIQ Labs' free, independent directory of AI agent skills plus a free MCP tool that connects any agent to the skills you choose. Unlike a marketplace that simply lists skills, SkillzBench verifies every skill (pinned to a commit SHA, validated, license-checked, security-scanned), benchmarks it in a sandbox with the same model with and without the skill to publish a Lift score in percentage points, and assigns a 0–100 trust score from security, quality, license, maintenance, and community signals. Everything is free, and advertising revenue never influences scores or listings. It is currently pre-launch, with the waitlist and newsletter live.

Which presentation skill produces real PowerPoint files instead of HTML decks?

SlideSpeak's agent skill produces native, editable .pptx files and offers both prompt-based and slide-by-slide generation modes designed for agents working from structured data like CRM exports. Felo Slides also exports .pptx files that opened cleanly in PowerPoint and Google Slides in benchmark testing, and Powerdrill Bloom's ppt-master generates downloadable PowerPoint files. Some other skills build polished HTML decks instead, so check the output format before committing to a workflow.

Are these skills free to use?

It varies. SkillzBench's directory, MCP connector, benchmarks, and alerts are all free. Felo Slides requires a Felo API key with a free tier available. SlideSpeak requires an API key. Several other tools, including Tosea.ai, Skywork, Agent Presentation Builder, and Powerdrill Bloom's skills, do not publish pricing in publicly available research, so you'll need to contact them or check their sites directly. Note that even where a skill itself is free, the underlying API or platform often is not.

Which agent platforms support presentation skills?

Skills using the SKILL.md format work across Claude Code, OpenClaw, Cursor, Codex, Gemini CLI, GitHub Copilot, Goose, Windsurf, Kiro, and other compatible agents — you typically install once and every agent in your environment gains the capability. Some options, like Powerdrill Bloom's ppt-master and mck-ppt-design, are platform-specific and only run inside their own workspace. OpenClaw users can also install skills through the ClawHub registry and trigger them from messaging apps like WhatsApp, Telegram, and Slack.

How do I know if a presentation skill actually improves my agent's output?

The honest answer is: benchmark it. The most reliable method is to run the same presentation task with and without the skill loaded, using the same model, and compare the results — which is exactly the methodology SkillzBench is building, publishing the difference as a Lift score in percentage points. Anthropic's own skill-creator skill similarly runs evals with and without a skill to measure impact. Without that comparison, install counts and marketing claims tell you little; a skill with 500,000 installs may still produce generic filler, as several high-install tools did in independent testing.