Stageshow Skills
Lets coding agents build React faux product UI and motion assets, then film the result to check its own work.
- Source repo
- benjitaylor/skills
- Stars
- ★ 3.9k
- Last updated
- 16d ago
- Primary language
- TypeScript
- FA score
- 45/100 · Major gaps
At a glance
- How it runs
- Works with
- Portable with changes
- Cost
- Free, no paid service needed
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- Marketing or product teams that need publishable imagery resembling the real product UI on short notice.
- Not a fit if
- Teams needing real shipped product UI rather than faux UI imagery
- Projects not on React or without a Node.js runtime
- Users wanting backend, testing or other non-design coding skills
- Maintenance
- Commits in the last 30 dayslast commit 2026-11-05latest release stageshow-v1.1.1 (2026-11-05)+87.6 ★/day over 7 days
- Source review
- 45/100 · Major gaps 1 safety controls not found
What does this agent do, and when should you use it?
This is a skills repository for AI coding agents, described as skills the author uses for design and motion work. The only publicly documented skill is Stageshow, which helps an agent produce faux product UI in React that resembles a real product: stills, feature cards, cursor demos and bento grids. The skill bundles composition and motion guidance, a kit to copy from, examples, and tooling that films the rendered result so the agent can verify its output rather than guess. Installation goes through the npx skills add command, with the skill body in stageshow/SKILL.md and an example preview in stageshow/README.md. The repository description is "Skills for AI coding agents"; the top-level license is unknown, and the README notes that each skill carries its own license and third-party asset notices.
Stageshow reads the composition and motion conventions in stageshow/SKILL.md, then generates React-based faux product UI following the included kit and examples — stills, feature cards, cursor demos and bento grids. It then uses the bundled filming tooling to capture the rendered output, so the agent can inspect screenshots or recordings and correct its own work instead of judging from source code alone. A preview section in stageshow/README.md shows what the examples look like before you commit. Everything runs in your local development environment and the output stays inside your own project.
- Marketing or product teams that need publishable imagery resembling the real product UI on short notice.
- Solo developers who want feature cards and bento grid illustrations for a landing page without drawing them by hand.
- Writers preparing product docs that need cursor-interaction stills to illustrate a flow.
- Frontend engineers on React projects who want coding agents to self-verify UI output instead of manual review.
Key setup facts
Pulled from the install and usage notes below. The project README has the current values.
- Install / run
npx skills add benjitaylor/skills --skill stageshow
How do you install or deploy this agent?
Install through the skills CLI into your current environment:
npx skills add benjitaylor/skills --skill stageshowThe repository does not document any alternative installation path (such as a manual git clone) or runtime prerequisites.
How do you use this agent?
After installation your coding agent reads the skill document and works from its guidance to produce React UI and filmed results:
stageshow/SKILL.mdExample output preview:
stageshow/README.md#previewA fuller explanation of how it works is at https://benji.org/stageshow. The repository gives no standalone command to run it without an agent.
What are this agent's strengths and limitations?
- Explicit output types: stills, feature cards, cursor demos and bento grids are each named.
- Ships a copyable kit and examples, so the agent is not designing from scratch.
- Includes filming tools that let the agent check rendered output instead of trusting code.
- One-line npx install keeps the adoption cost low.
- License is listed as unknown; the README only says each skill carries its own, so you must check stageshow's terms yourself.
- Only one skill is documented, so the broad "Skills" framing overstates current coverage.
- Docs are written for agent consumption, leaving humans short on runtime and dependency details.
- Output assumes React; support for other frontend frameworks is not stated.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Stageshow Skills This agent | 45 · Major gaps | Agent plugin / skillFree | ★ 3.9k | 16d ago | TypeScript | — |
| UIZZE · Better UI for Coding Agents | 69 · Some gaps | Agent plugin / skillFreemium | ★ 121 | 21d ago | JavaScript | Codex · Claude Code |
| Appllama Mobile Design Skills | 64 · Some gaps | Agent plugin / skillFreemium | ★ 5.3k | 2mo ago | — | Codex · Claude Code |
| Video Shotcraft | 74 · Some gaps | Agent plugin / skillFree | ★ 22k | 2d ago | TypeScript | Codex · Claude Code |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows a local React illustration toolkit: brands/index.ts explicitly states 'No network lookup or runtime icon service', and SKILL.md frames output as authored illustrations rather than browser automation, so external effects and exfiltration risk are low. However this is low risk rather than an explicit permission posture: there is no permission manifest and no stated boundary on file writes or command execution, so least_privilege is only 1. user_confirmation: SKILL.md mentions filming results via scripts/film.mjs but never says whether user confirmation is required or where artifacts land, 1. data_flow_transparency is only scattered comments, 1. sensitive_data_handling has no policy; scored 1 only because no sensitive data is collected. dependency_security: no lockfile or audit note in the packet, 1. external_effects: local rendering, 1. rollback is absent: no undo, artifact cleanup, or failure recovery, 0. source_attribution has real evidence: README states each skill includes its own licence and third-party asset notices, and brands/ annotates each mark with light/dark variants, 2.
self_consistency is good: AGENTS.md's Node 20.19 requirement, the typecheck/check:css checks, and SKILL.md's version 1.1.1 align with the kit export structure, 2. dependency_availability: only npx skills add and npm install are mentioned; no package.json or lockfile evidence, so availability is unconfirmed, 1. failure_messages: no error handling or diagnostics in the sampled source and no failure guidance in SKILL.md, 1.
audience_and_scenarios is clear: design/motion work for coding agents, with SKILL.md detailing hero, card, bento and phone-column sizes, 2. capability_boundaries is explicit: 'not browser automation', a.click()/a.set() draw rather than interact, plus five named failure modes, 2. trigger_precision: the description covers many phrasings including 'even if they never say faux UI', 2. environment_fit: only Node 20.19+ and React/TypeScript/CSS are stated; host-agent and bundler differences are acknowledged only in a kit comment, 1.
information_architecture is solid: each skill is a self-contained folder and the kit is layered into stage/frames/motion/parts/controls with index comments, 2. install_notes gives both npx skills add and local npm install/dev, 2. naming_stability: semantic version 1.1.1 and stable export names, 2. examples_and_faq: gold examples in examples/ plus a README preview link, 2. known_limitations: only scattered remarks (e.g. style removal depends on the host bundler), no consolidated section, 1. license: only a README sentence that each skill carries its own licence; package-level licence metadata is unknown, 1. versioning_changelog: only metadata.version, no changelog, 1. maintenance_responsibility: publisher unverified and no maintainer or support channel documented, 1.
output_usability is strong: output is usable React components plus a filmable demo, with a stepwise workflow, 2. marginal_value is differentiated: it encodes composition and motion rules specific to faux product UI rather than generic code generation, 2. cost_benefit: requires dependency install and a filming script, with no performance or size evidence, so benefit evidence is limited, 1.
claim_traceability is partial: README links to a benji.org article and the skill doc, but quality claims ('gold examples are the bar') have no measurable basis, 1. cross_source_corroboration is limited: README, AGENTS.md and SKILL.md agree on Node version and workflow, but core quality claims rest on a single source, 1. fact_inference_separation is mixed: SKILL.md blends design judgement with factual description without marking which is which, 1.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- Publisher identity is unverified and repository-level licence metadata is unknown; README only says each skill carries its own licence, so confirm per-skill licensing and rights to bundled third-party brand assets before commercial use.
- The scripts/film.mjs filming flow referenced in SKILL.md does not state where artifacts are written, whether existing files are overwritten, or whether user confirmation is required; inspect the script before running it.
- No package.json or lockfile appears in the packet, so dependency availability and version pinning cannot be confirmed statically.
- The brands directory bundles many third-party trademarks; using generated output in public marketing material carries trademark-compliance risk.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
stageshow-v1.1.1 Stageshow 1.1.1
Release notes excerpt
Fixes the playground shaking at some widths when scroll bars are always shown (#6), including the same loop in the viewer page itself.View on GitHub ↗ -
stageshow-v1.1.0 Stageshow 1.1.0
Release notes excerpt
- New kit parts: Button, Card (with a flush option for sectioned cards), Divider, Kbd, and Draft, a comment being typed that becomes the posted comment in place. - Smoother Collapse: content stays on screen while it closes, and reveal="drop" glides even with a demo's own transitions. - Smoother Roll: text rolls under a soft edge instead of a hard clip, and the old and new words move together, so the line never goes blank. - Stricter rehearsal: always film with --gate, and don't hand over until it passes. The gate now exits 1 when it fails (#5). …View on GitHub ↗
FAQ
Does using this skill cost anything?
Will it change files in my project?
Is the output a real product interface?
Are there skills besides Stageshow?
Is Stageshow Skills open source?
Is Stageshow Skills safe to use?
Is Stageshow Skills still maintained?
Who makes Stageshow Skills?
Thanks — we use this to improve the page.