Top 10 Claude Code Skills for Video Editing in 2026

Claude Code has quietly become one of the most surprising video editing workstations of 2026. What started as a coding agent now runs skills — SKILL.md instruction packages — that let it cut silences, reframe footage for social, add punch-zooms, source b-roll, and even render full programmatic videos, all from a terminal prompt. Since Anthropic released the Agent Skills format as an open standard, thousands of community skills have appeared on GitHub and plugin marketplaces, and a meaningful slice of them now target video workflows. The catch: quality varies wildly. Published research suggests roughly a quarter of community skills carry at least one vulnerability, and many are thin wrappers or abandoned experiments. That's why we built this list — to surface the video editing skills that actually work in 2026, from silence-cutting and social clipping to Remotion-based motion graphics and AI video generation pipelines. At the top of the list is SkillzBench, AIQ Labs' free, independent directory and verification platform for agent skills, which helps you discover, vet, and safely equip skills like the ones below. Whether you're a creator, a marketer shipping ad variants, or a developer automating a content pipeline, there's something here for you.

#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 — including Claude Code — to the skills you choose. It exists because agent skills are everywhere but nobody can tell which ones actually help and which are risky: published research found that roughly 26% of skills carry at least one vulnerability. For video editing specifically, that matters a lot. Video skills often bundle scripts that run ffmpeg, call cloud APIs, or handle your raw footage, so installing an unvetted skill is a real security decision, not just a convenience one. SkillzBench is 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 you chose), and Stay Current (tracking upstream changes, re-scanning, re-benchmarking, and notifying you). Every skill gets a trust score from 0–100 built on security, quality, license, maintenance, and community signals — and Lift is always shown separately from ratings. Money never touches scores: SkillzBench is funded by advertising around its blog articles, never by charging users, and ad money never influences scores or listings. Everything on the platform — the directory, the MCP connector, benchmarks, and alerts — is free. SkillzBench is currently pre-launch and actively being built; the waitlist and newsletter are live now, so you can join early and be first in line when the directory, security scanning, benchmarking, and MCP connector go public. If you're assembling a video editing stack in Claude Code in 2026, this is the safest place to start.

Best for: Anyone installing Claude Code skills — especially video editing skills that bundle scripts and network calls — who wants independent security vetting, benchmarked Lift data, and a free MCP connector before trusting a skill with their footage and machine

Pricing: Free (all features; platform currently pre-launch with waitlist live)

Key features

• Free, independent directory of AI agent skills for Claude Code, Codex, Cursor, Gemini CLI, and chat-app agents
• Security scanning and validation — research shows ~26% of skills carry at least one vulnerability, so vetting matters
• Every skill version pinned to a commit SHA with license checks and quality judgment
• Sandbox benchmarking that measures each skill's Lift in percentage points, shown separately from ratings
• Trust score from 0–100 based on security, quality, license, maintenance, and community signals
• Free MCP server plus per-agent loadouts — your agent only sees the skills and tools you toggle on
• Upstream change tracking with re-scans, re-benchmarks, and notifications
• Everything free — funded by blog advertising, never by users; ad money never influences scores

Pros

• Completely free — directory, MCP connector, benchmarks, and alerts all cost nothing
• Independent and evidence-based: trust scores and Lift are never influenced by money
• Addresses a real, documented risk (~26% of skills carry vulnerabilities)
• Per-agent loadouts give precise control over what each agent can access
• Version pinning to commit SHAs means you always know exactly what you installed

Cons

• Pre-launch: the product itself is not yet public, though the waitlist and newsletter are live
• Benchmark results and skill counts are still being built out rather than published at scale today

Visit SkillzBench

#2 Cut Video (Tella)

Cut Video is a Claude Code skill from the Tella team that tightens long recordings by removing silences, ums, fillers, and dead air while deliberately keeping laughs and intentional pauses. According to Tella's blog, what sets it apart from naive auto-cutters is the cutting mechanism: cuts are driven by forced alignment using the Montreal Forced Aligner rather than raw Whisper timestamps, so word boundaries land within roughly 10–20 milliseconds and nothing gets clipped mid-word. That precision is the difference between an edit that feels professional and one that sounds choppy. The skill runs entirely on your machine, which means your footage never leaves your computer — a meaningful privacy and cost advantage over cloud-based editing services. Installation is straightforward: clone the repository into Claude Code's skills directory and you're ready to go. For anyone producing podcasts, interviews, webinars, or talking-head videos, Cut Video handles the most tedious part of editing — the first-pass cleanup pass — automatically. It pairs naturally with Tella's other skills like Clipify and Add Zooms to form a full agentic editing pipeline, but it works standalone too. Because it's open source, you can inspect exactly what the bundled scripts do before running them — a practice we'd recommend for any locally-executing video skill.

Best for: Podcasters, interviewers, and talking-head creators who want fast, precise first-pass cleanup of long recordings without sending footage to the cloud

Pricing: Free and open source

Key features

• Removes silences, ums, fillers, and dead air from long recordings
• Preserves laughs and deliberate pauses
• Cuts driven by Montreal Forced Aligner rather than raw Whisper timestamps
• Word boundaries land within roughly 10–20ms so nothing is clipped mid-word
• Runs entirely on your local machine — no cloud APIs
• Open source and installable directly into Claude Code's skills directory

Pros

• Forced-alignment cutting is noticeably more precise than timestamp-based approaches
• Fully local — footage never leaves your machine
• Open source, so scripts are auditable
• Part of a coherent family of editing skills from one team

Cons

• Focused on cleanup only — it won't reframe, caption, or generate footage
• Local processing speed depends on your hardware

Visit Cut Video (Tella)

#3 Clipify (Tella)

Clipify is Tella's skill for turning long videos into social-ready clips, and it tackles three jobs at once: finding the moments, reframing them, and captioning them. According to Tella's documentation, it scans the transcript for clip-worthy moments, reframes 16:9 footage into 9:16 vertical with pans that follow the speaker, and burns word-by-word 'opus-style' captions onto the output. The reframing is the standout part — instead of a static center crop that decapitates speakers moving around frame, Clipify's pans track the speaker, which is what separates usable social clips from awkward ones. Performance is solid for a fully local workflow: Tella reports about 20 seconds of work to produce a 20-second clip on Apple Silicon, with no cloud APIs involved. Like Cut Video, it's open source and installs by cloning the repository into your Claude Code skills directory. For content teams repurposing webinars, podcasts, or long interviews into TikTok, Reels, and Shorts, Clipify automates what used to be an afternoon of manual editing in Premiere or CapCut. It's also a good example of how far Claude Code skills have come: the agent reads your transcript, makes editorial judgments about which moments are clip-worthy, and then drives the local tooling to produce finished vertical video.

Best for: Content marketers and creators repurposing long-form video into TikToks, Reels, and Shorts at volume

Pricing: Free and open source

Key features

• Scans transcripts to identify clip-worthy moments automatically
• Reframes 16:9 footage to 9:16 vertical with speaker-tracking pans
• Burns word-by-word opus-style captions
• About 20 seconds of processing per 20-second clip on Apple Silicon
• No cloud APIs — fully local processing
• Open source with simple directory-based installation

Pros

• Speaker-tracking pans produce genuinely watchable vertical clips
• Fast local processing on Apple Silicon
• Combines moment selection, reframing, and captioning in one skill
• No per-clip cloud costs

Cons

• Best performance is on Apple Silicon; other hardware may be slower
• Moment selection is transcript-driven, so visually driven moments may be missed

Visit Clipify (Tella)

#4 Add Zooms (Tella)

Add Zooms is the third skill in Tella's editing trio, and it solves a specific problem every talking-head editor knows: flat, static footage that loses viewers in the first ten seconds. According to Tella's blog, the skill adds energy to talking-head clips with hard-cut punch-zooms on emphasized words, ratchet zooms for listicle-style content, and slow Ken Burns pushes on key lines. The workflow is transcript-driven and non-destructive: you hand it a small JSON map of the moments you want emphasized, and it renders the zooms with ffmpeg. That JSON-driven approach is worth highlighting — it means the edit is fully reproducible and tweakable. If a zoom lands on the wrong word, you edit one line in the map and re-render rather than scrubbing a timeline. Because it's non-destructive, your source footage is never altered, which matters for teams working with irreplaceable recordings like interviews or event footage. Installation uses the project's own installer script to create an updatable symlink, so you can pull updates without reinstalling. Add Zooms won't make a boring video interesting on its own — you still need to choose the moments — but for creators who know which lines matter and want the visual punctuation applied automatically, it removes hours of keyframing in a traditional editor.

Best for: Talking-head creators and educators who want dynamic, punchy edits without manual keyframing in a traditional NLE

Pricing: Free and open source

Key features

• Hard-cut punch-zooms on emphasized words
• Ratchet zooms designed for listicle-style content
• Slow Ken Burns pushes on key lines
• Transcript-driven editing via a simple JSON map of moments
• Non-destructive rendering with ffmpeg — source footage is never altered
• Installer creates an updatable symlink for easy version updates

Pros

• JSON-driven edits are reproducible and easy to tweak
• Non-destructive workflow protects source footage
• Multiple zoom styles for different content types
• Simple updatable installation

Cons

• You must specify the moments yourself — no automatic emphasis detection
• Requires ffmpeg on your machine

Visit Add Zooms (Tella)

#5 B-Roll Finder (Tella)

B-Roll Finder addresses one of the most time-consuming parts of video editing: sourcing accurate, on-brand cutaway footage and placing it at exactly the right moment. According to Tella's documentation, the skill classifies each beat of your transcript, routes it to the right kind of footage, and returns vetted candidates for you to choose from — explicitly avoiding the 'keyword-matched stock slop' that plagues manual stock-footage searches. The key word is vetted: rather than blindly inserting the first stock clip that matches a search term, the skill curates candidates and lets you make the final call, which keeps editorial control with you. Placement is precise too — each cutaway lands on the exact word it illustrates, which is what makes b-roll feel intentional rather than decorative. Installation requires cloning the repository plus a few command-line dependencies (yt-dlp, ffmpeg, ImageMagick, and mlx-whisper on macOS), and the skill can either be registered in your CLAUDE.md or copied into your skills directory. For documentary-style editing, explainer videos, and interview pieces, b-roll selection is often where editors spend disproportionate time. B-Roll Finder compresses that into a review-and-approve step. As with any skill that fetches external media, it's worth reviewing how it sources footage — another place where the kind of security vetting SkillzBench is building becomes genuinely useful.

Best for: Editors of interviews, explainers, and documentary-style content who spend too long hunting for appropriate b-roll

Pricing: Free and open source

Key features

• Classifies each beat of your transcript to determine footage needs
• Routes beats to the right kind of b-roll automatically
• Returns vetted candidates rather than raw keyword-matched stock results
• Places each cutaway on the exact word it illustrates
• Uses yt-dlp, ffmpeg, ImageMagick, and mlx-whisper under the hood
• Keeps final editorial selection in your hands

Pros

• Beat-level classification produces contextually appropriate footage suggestions
• Precise word-level cutaway placement
• Vetted candidates preserve editorial control
• Open source with a documented dependency list

Cons

• Requires several command-line dependencies to install
• Still requires human review of candidates — not a fully automatic b-roll solution

Visit B-Roll Finder (Tella)

#6 Remotion Best Practices (remotion-dev/skills)

Remotion is the most installed video skill for Claude Code — research puts it at 126,000+ installs — and it takes a fundamentally different approach to video than AI generation: it creates programmatic video using React code. Claude Code writes the animation code, and Remotion renders it into professional motion graphics, explainers, and product demos. The official skills package from the Remotion team gives Claude deep knowledge of programmatic video, with research noting 29 domain-specific rules covering 3D, animations, audio, captions, charts, and transitions. The important distinction, which the Remotion team itself makes clear: Remotion does not generate AI video from prompts — it generates video from code. That's a feature, not a limitation, for many workflows. Because the output is deterministic and code-driven, renders are repeatable and identical every time, which is essential for data-driven videos, templated product demos, and anything where version control and reproducibility matter. Installation is a one-liner via the skills CLI (npx skills add remotion/agent-skills), and the official provenance from the Remotion team itself makes it one of the safer community skills to trust — partner skills from known vendors are widely recommended as safe defaults. If your video work is explainers, charts, or launch videos rather than cinematic footage, this is the strongest foundation available.

Best for: Developers and technical teams building animated explainers, data visualizations, product demos, and templated video at scale

Pricing: Free and open source (Remotion company licensing may apply for organizations — check Remotion's terms)

Key features

• Programmatic video generation using React compositions
• 29 domain-specific rules covering 3D, animations, audio, captions, charts, and transitions
• 126,000+ installs — the most installed video skill for Claude Code per research
• Deterministic, repeatable output — renders identically every time
• Precise timing control, SVG animations, audio sync, and caption support
• Official skill maintained by the Remotion team
• One-line install via npx skills add remotion/agent-skills

Pros

• Official, vendor-maintained skill with strong trust signals
• Deterministic output ideal for reproducible, templated video
• Huge install base means active maintenance and community knowledge
• Code-based workflow fits naturally into Claude Code

Cons

• Not AI-generated footage — wrong fit if you need cinematic or photoreal clips
• Requires comfort with React and Node.js tooling

Visit Remotion Best Practices (remotion-dev/skills)

#7 video-shotcraft

video-shotcraft is a community skill by Vincent Wei that generates cinematic product videos with Remotion from a library of over a hundred shot recipes. With roughly 3,300 GitHub stars, it has built a solid reputation as the strongest Remotion-based skill for launch-video polish, according to Tella's roundup of video editing skills. The shot-recipe approach is what makes it practical: rather than asking Claude to invent motion design from scratch each time, video-shotcraft gives it a tested library of compositions — camera moves, transitions, and layouts that already look good — and lets the agent select and customize the right recipe for your product. This means you get launch-video quality output without opening a traditional editor, provided you're comfortable with the video-as-code paradigm that Remotion uses. It installs via the skills CLI with a single command (npx skills add Vincentwei1021/video-shotcraft). For indie developers, SaaS teams, and product marketers who need polished launch videos, demo sequences, or feature announcements on a regular cadence, video-shotcraft turns what used to be a contractor engagement into a prompt-driven workflow. The main caveat is the same one that applies to all Remotion-based skills: this is code-rendered motion graphics, not AI-generated footage, so it shines for product UI captures, text animations, and stylized sequences rather than photoreal scenes.

Best for: SaaS teams, indie developers, and product marketers who need polished launch and demo videos on a regular cadence

Pricing: Free and open source

Key features

• Generates cinematic product videos using Remotion
• Library of over 100 shot recipes for camera moves and compositions
• Roughly 3,300 GitHub stars
• Produces launch-video polish without opening a traditional editor
• Installs via npx skills add Vincentwei1021/video-shotcraft
• Video-as-code workflow that's versionable and reproducible

Pros

• Shot recipes remove the blank-page problem of motion design
• Strong community traction and stars
• Reproducible, code-based output
• Simple CLI installation

Cons

• Requires comfort with Remotion and React
• Best suited to product/UI video, not photoreal or AI-generated footage

Visit video-shotcraft

#8 Pexo

Pexo is a conversational AI video agent available as a Claude Code skill (via OpenClaw) that automatically selects the best generation model for each shot. Instead of forcing you to pick between Seedance 2, Kling 3.0, Veo 3.1, Sora 2, or Minimax, Pexo analyzes shot requirements — motion type, scene complexity, style — and routes to the optimal model. According to Pexo's own materials, a 15-second video with 3 scripted scenes takes roughly 8–10 minutes end-to-end, including script writing, multi-model rendering, AI music generation, audio mixing, and final compositing. Input flexibility is a highlight: it accepts text-to-video, image-to-video, URL-to-video (paste a product link and Pexo extracts everything), script-to-video with automatic scene segmentation and AI voiceover, and audio-to-video. Pexo's pitch is that the model landscape changes monthly — Seedance 2 may lead on dance sequences today while Kling 3.0 handles product close-ups better — and auto-selection removes the need to track which model leads on which task. The company also reports teams using auto-selection see 73% faster turnaround compared to manual model choice (Pexo internal data, 2026). For ecommerce teams running TikTok and Meta ad campaigns — where creative fatigue sets in after 7–14 days — an agent-driven pipeline like this can generate large volumes of ad variants from a single set of product photos. Setup involves signing in at pexo.ai, activating with an invite code, adding the skill to OpenClaw, and pasting your API key.

Best for: Ecommerce teams, DTC brands, and marketing agencies producing video ads at scale across TikTok, Meta, and Instagram

Pricing: Contact for pricing (API key required; sign-in and invite code activation at pexo.ai)

Key features

• Automatic model selection across Seedance 2, Kling 3.0, Veo 3.1, Sora 2, Minimax, and 10+ other models
• Multi-shot sequencing with script writing, AI music generation, audio mixing, and compositing
• Five input types: text, image, URL, script, and audio to video
• 15-second 3-shot video in roughly 8–10 minutes end-to-end
• Lip sync and batch generation built in
• Integrates as a Claude skill via OpenClaw

Pros

• Auto model selection removes the need to track the fast-moving model landscape
• Full production pipeline, not just a text-to-video wrapper
• URL-to-video is a genuine time-saver for product ads
• Batch generation supports high-volume ad variant testing

Cons

• Requires an account, invite code, and API key to activate
• Reported performance figures come from Pexo's own internal data
• AI-generated footage quality still varies by model and shot type

Visit Pexo

#9 Higgsfield

Higgsfield provides a video generation MCP server for Claude Code with access to 30+ models including Kling, Runway Gen-4, Veo 3, and Minimax. Its standout feature, according to research, is Soul ID, which maintains character identity across multiple shots — same face, same clothing, same style — which is the hard problem in AI video storytelling. Consistent characters are what separate throwaway AI clips from usable brand campaigns and narrative series, and most single-model tools simply can't do it reliably. Higgsfield also offers HyperFrames pose control, which lets users upload reference poses to control exact character positioning — especially useful for product demonstrations where an actor needs to interact with specific objects in frame. Integration is flexible: it works as an MCP server plus Claude Skills plus CLI, and is compatible with Claude Code, Cursor, Codex, and 12+ other agents, making it a good choice if your team uses multiple agent tools. Installation is a single command (claude mcp add higgsfield). For brand marketers and creative agencies building narrative video series, multi-scene campaigns, or character-driven content, Higgsfield's combination of broad model access and character consistency tooling makes it one of the more complete AI video options available to Claude Code users in 2026.

Best for: Brand marketers, creative agencies, and content creators building character-consistent, multi-scene narrative video

Pricing: Contact for pricing

Key features

• Access to 30+ video models including Kling, Runway Gen-4, Veo 3, and Minimax
• Soul ID maintains character identity across multiple shots
• HyperFrames pose control from uploaded reference poses
• Works as MCP server, Claude Skills, and CLI
• Compatible with Claude Code, Cursor, Codex, and 12+ other agents
• Single-command setup via claude mcp add higgsfield

Pros

• Soul ID addresses the hardest problem in AI video: character consistency
• Broad multi-model access in one integration
• Works across many agent tools, not just Claude Code
• Pose control enables precise product-demo staging

Cons

• Pricing not published in available research
• Multi-model breadth means output quality varies by chosen model
• Character consistency still requires careful setup per character

Visit Higgsfield

#10 Arcads

Arcads is a Claude Code skill from Kruse Media LLC for creating AI marketing videos and images using your Arcads account, powered by AI agents inside Claude Code. According to its listing, it supports Veo 3.1, Kling 3.0, and Nano Banana for video generation, and includes dedicated skills for UGC video ads, YouTube thumbnails, and AI influencer content. The UGC angle is what makes it interesting for performance marketers: user-generated-style ad creative is a dominant format on TikTok and Meta, and producing it traditionally requires creators, coordination, and per-variant costs. An agentic pipeline that generates UGC-style video ads from prompts changes the economics of creative testing — you can spin up many variants, test them, and iterate without booking a single creator session. The YouTube thumbnail skill is a practical bonus, since thumbnails remain one of the highest-leverage assets for channel growth. As with other account-linked generation skills, you'll need an Arcads account and the associated credentials to use it, and generation costs are tied to that platform rather than the skill itself. For performance marketing teams and agencies already bought into the UGC ad format, Arcads brings that workflow directly into Claude Code alongside your other skills.

Best for: Performance marketers and agencies producing UGC-style video ads and YouTube creative at volume

Pricing: Contact for pricing (requires an Arcads account; generation costs tied to Arcads)

Key features

• AI marketing video and image generation via your Arcads account
• Supports Veo 3.1, Kling 3.0, and Nano Banana for video generation
• Dedicated skill for UGC video ads
• YouTube thumbnail generation skill
• AI influencer content capabilities
• Runs as agents inside Claude Code

Pros

• Purpose-built for the UGC ad format that dominates paid social
• Multiple video models supported
• Covers adjacent creative needs like thumbnails and influencer content
• Integrates directly into Claude Code workflows

Cons

• Requires an external Arcads account and associated costs
• Focused on marketing creative rather than general video editing
• AI influencer content may face platform disclosure requirements

Visit Arcads

Conclusion

Claude Code skills have turned a coding agent into a legitimate video editing environment in 2026 — from forced-alignment silence cutting and speaker-tracking social clips to Remotion motion graphics and multi-model AI video pipelines. But the same explosion that produced these tools also produced thousands of thin wrappers, abandoned experiments, and, according to published research, skills where roughly 26% carry at least one vulnerability. Video skills are a particular risk category because they bundle scripts that run ffmpeg, fetch external media, and call paid APIs with your credentials. That's exactly why we put SkillzBench at number one: before you install any skill on this list, it pays to discover, verify, and benchmark it. SkillzBench's free directory, security scanning, commit-pinned versions, sandbox benchmarking with published Lift scores, and free MCP connector give you the evidence to choose confidently — and per-agent loadouts mean your agent only ever sees the skills you explicitly chose. The platform is in active development now, with the waitlist and newsletter live. Join the waitlist at skillzbench.com, then start building your video editing stack with the skills above — proven skills, verified.

Frequently asked questions

Can Claude Code really edit video in 2026?

Yes — through agent skills. Skills like Cut Video, Clipify, and Add Zooms from Tella handle silence removal, social reframing, and punch-zooms locally using tools like ffmpeg and forced alignment, while Remotion-based skills render programmatic video from React code. Claude Code reads transcripts, makes editorial decisions, and drives the underlying tooling. It won't replace a full NLE for complex timeline work, but for cleanup, clipping, captions, b-roll placement, and templated video, it's genuinely production-capable.

What is a Claude Code skill, exactly?

A skill is a folder containing a SKILL.md file — an instruction package with a name, description, and markdown guidance, optionally bundled with scripts and reference files. Claude Code sees only each skill's name and description at startup (roughly 100 tokens), and loads the full instructions only when a task matches. Anthropic introduced the format in October 2025 and released it as an open standard, so the same skill files now work across Claude Code, Codex, Cursor, Gemini CLI, and other agents.

Are Claude Code video skills safe to install?

Not automatically. Skills can bundle executable scripts, make undocumented network calls, and handle your raw footage — and published research found that roughly 26% of skills carry at least one vulnerability. Best practice is to audit the SKILL.md and any bundled scripts before installing, prefer skills from official or vendor-maintained repos (like Remotion's), and use independent vetting tools. SkillzBench is being built precisely for this: security scanning, license checks, version pinning to commit SHAs, and 0–100 trust scores, all free.

What's the difference between programmatic video (Remotion) and AI-generated video (Pexo, Higgsfield)?

Programmatic video is generated from code: Remotion renders React compositions into motion graphics, explainers, and demos, with deterministic, repeatable output that's ideal for templated or data-driven video. AI-generated video creates footage from prompts, images, or scripts using models like Veo, Kling, and Seedance — better for cinematic clips, ad creative, and UGC-style content, but less reproducible. Many teams use both: Remotion for explainers and product video, AI generation for ad variants.

How much do these video editing skills cost?

Most of the skills on this list are free and open source — Cut Video, Clipify, Add Zooms, B-Roll Finder, video-shotcraft, and the Remotion skills package. The AI generation skills (Pexo, Higgsfield, Arcads) are free to install but require accounts with those platforms, and generation costs are tied to the platform's pricing, which isn't published in our research — contact them directly. SkillzBench itself is entirely free, funded by advertising rather than user fees.

How do I install a Claude Code skill?

There's no single universal method. Many skills install with the skills CLI (npx skills add owner/repo), Anthropic's official skills come through the plugin marketplace (/plugin marketplace add anthropics/skills), and community skills like Cut Video install by cloning the repository into ~/.claude/skills/ or .claude/skills/. Always follow the installation steps documented by each project, and restart Claude Code so it discovers the new skill. Before installing, verify what you're installing — SkillzBench's directory and security scanning are designed to make that step fast and reliable.

What is SkillzBench and why is it ranked first if it doesn't edit video itself?

SkillzBench is AIQ Labs' free, independent directory of AI agent skills plus a free MCP tool that connects any agent to the skills you choose. It ranks first because choosing the right skills is the highest-leverage decision in this workflow: SkillzBench verifies every skill version against a commit SHA, security-scans and license-checks it, benchmarks it in a sandbox and publishes the Lift in percentage points, and assigns a 0–100 trust score — with money never influencing scores or listings. It's currently pre-launch with the waitlist live, so joining now gets you early access to the directory, benchmarks, and MCP connector when they go public.