Vellum Assistant is a personal AI assistant that remembers information about users, learns their preferences, and takes actions across connected apps. It is intended for people who want an assistant that can manage conversations, unfinished work, and proactive notifications over time. The catalogue skills, hooks, instruction, and setting configure or extend how the assistant works.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add vellum-ai/vellum-assistant --skill visualizegit clone --depth 1 https://github.com/vellum-ai/vellum-assistantWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/vellum-ai/vellum-assistant/visualize)<a href="https://agentmods.dev/skills/vellum-ai/vellum-assistant/visualize"><img src="https://agentmods.dev/badge/skills/vellum-ai/vellum-assistant/visualize/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/vellum-ai/vellum-assistant/visualize"><img src="https://agentmods.dev/badge/skills/vellum-ai/vellum-assistant/visualize.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00057 | $0.03536 |
| Opus 5.5 | $0.00023 | $0.01414 |
| Sonnet 5.5 | $0.00011 | $0.00707 |
| Haiku 4.5 | $0.00006 | $0.00354 |
Grade A, and why
visualize scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are authoring a self-contained HTML fragment that renders inline in the chat transcript, directly beneath the paragraph you are writing. It renders in a sandboxed frame sized to your content, with the host's design tokens injected, so it looks native in light and dark mode.
Invocation
ui_show { surface_type: "visual", data: { html: "<fragment>", height: 320 } }
heightis a rough pixel estimate (80 to 1400); the host measures and corrects after first paint.- One visual per call. To change one already shown,
ui_dismissitssurface_idand show a new one. - Prose goes in your reply, the visual goes in the tool: no titles, intros, or captions inside the fragment, and never describe in prose what the visual already shows.
- If ui_show returns an error, read it, fix the arguments, and call again. Every rejection is fixable in the next call; never debug via shell or files, and never narrate retries.
A visual is part of one answer. Durable things the user reopens (a dashboard, a tracker, a calculator they keep) are apps: use the app-builder skill.
Design instincts
- Seamless and flat: no gradients, shadows, glow, or texture. Flat fills, hairline borders.
- Compact and honest: show the essential, explain the rest in prose; every number on screen is one you actually have.
- Pick the lightest form that carries the idea, and invent freely: a bespoke drawing that fits this answer beats a stock layout. Interaction has to earn its place; steppers are for stages that genuinely follow one another, tabs are not a default. Vary layout between visuals in one conversation.
- Route diagrams on the verb: "walk me through" wants a flowchart; "how is it organised" wants containment boxes; "how does it actually work" wants the mechanism drawn as an intuition, not a safer flowchart.
Sandbox constraints (hard)
- No network: no script src, stylesheets, @import, remote images, web fonts, fetch, or CDN libraries. Charts and diagrams are hand-drawn inline SVG; images are inline SVG only.
- Fragment only: no DOCTYPE, html, head, or body. No HTML comments, no emoji.
- Hard cap 24000 characters; aim well under 8000. Too big to fit is too big to render. Split rich subjects into two visuals with prose between.
- No position fixed or sticky, no nested scrollbars: the frame auto-sizes from in-flow content.
- No localStorage, sessionStorage, or cookies. Hold state in JS variables.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 164 lines · 57 tokens per session scan A f001c78cd272
visualize is a skill published in the GitHub repository vellum-ai/vellum-assistant (1,448 stars, last pushed 2d ago), licensed MIT. It adds 57 tokens to every session and 3,536 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
html-artifacts
Author the HTML for a plan artifact, dashboard iframe, or Slack attachment — structure, design plan, available runtime, theming, and craft. Read this before writing HTML for saveplan, outputiframe, or slackattachhtml.
mermaid-diagram
Use when the user asks for a flowchart, architecture diagram, sequence diagram, ER diagram, state machine, gantt chart, or any visual that fits Mermaid syntax. Hope Agent's chat renders Mermaid blocks natively, so the diagram appears inline.
design-review
Use when a design artifact or revision needs a prepared review packet or an evidence-backed record of a review outcome, requirements, and decisions.
generative-ui
Use when presenting information already in hand as a small, static structured panel in the conversation — status, results, comparisons, or a compact summary — rather than prose, a markdown table, a widget, or an app.
st-widget-builder
Use when the user asks to render, build, visualise, mock up, or show a widget (chart, dashboard, card, table, graph) in the Mewbo Console. Teaches how to delegate the work to the st-widget-builder sub-agent instead of writing HTML or inline code yourself.
frame-composition
Stage the frame on a NodeTool timeline — grids, focal placement, safe areas per aspect ratio, depth layers and parallax, camera moves, and where elements enter and leave. Use when placing titles, lower thirds, shapes or overlays, when a layout must survive 16:9 and 9:16, when adding a push or a parallax move, or when…