# Answer Streams In — effect skill for AI agent flows

> **For AI agents:** find the right skill in one call — `curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>"` returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add `.md` to its path. Before generating anything, show the person the options on the how-to page.

The assistant's reply arrives in uneven token bursts, the way a real model streams.

- **Type:** [Effect](https://varg.ai/how-to/effects.md) · [AI agent flows](https://varg.ai/how-to/browse/skills/agent-flows.md)
- **Slug:** `answer-streams-in`
- **Status:** live
- **Publisher:** Miao Yang · HeyGen (community)
- **License:** Apache-2.0
- **Source:** https://github.com/heygen-com/hyperframes/tree/main/registry/components/streaming-text
- **Works in:** Claude Code, Claude.ai, Codex, Cursor, OpenClaw, Gemini CLI, Any MCP client, varg Studio
- **Requires:** HyperFrames (npx hyperframes), Remotion, CapCut (by hand)
- **Cost:** Free — the take renders locally; generated footage, if any, bills by the model

## What it does

Built on [Streaming Text](https://github.com/heygen-com/hyperframes/tree/main/registry/components/streaming-text) by Miao Yang (hyperframes), Apache-2.0. The write-up and recipes below are varg's.

Words appear in short irregular bursts with a soft caret; markdown headings and lists form as they stream.

**When:** Chat assistants and any feature whose output is text.

**Watch out:** Do not reveal at a perfectly even rate; real streams come in bursts and that rhythm sells authenticity.

**Timing:** Bursts of 2–6 tokens every 40–120 ms; total ≤ 3 s on screen, then hold.

**Also called:** streaming text, token stream, AI reply

### In HyperFrames

```bash
npx hyperframes add streaming-text
```
Use a fixed seed for burst sizes so every render streams identically.

### In Remotion

```tsx
const bursts = useMemo(()=>seededBursts(text, 42),[]); // [{at,frame,upto}]
const upto = bursts.filter(b=>b.frame<=frame).at(-1)?.upto ?? 0;
<Markdown>{text.slice(0,upto)}</Markdown>
```

### In CapCut

Record the real stream; or Text → Animation In "Typewriter" with a long duration on a pasted answer.

### Ready-made components

- [Streaming Text](https://skills.varg.ai/component/hf.streaming-text/) (hyperframes): npx hyperframes add streaming-text
- [Claude Exchange](https://skills.varg.ai/component/hf.claude-exchange/) (hyperframes): npx hyperframes add claude-exchange
- [Text Generate Effect](https://skills.varg.ai/component/aceternity.text-generate-effect/) (aceternity): npx shadcn@latest add @aceternity/text-generate-effect

### Where this comes from

- [Glama: designing tool-enabled chat interfaces](https://glama.ai/blog/2025-08-27-designing-tool-enabled-chat-interfaces)
- [Trace visualization: showing users what your AI agent is doing](https://agentinterface-co.pages.dev/blog/trace-visualization-showing-users-what-ai-agent-does.html)

## Use when

- Chat assistants and any feature whose output is text.

## Not for

- Do not reveal at a perfectly even rate; real streams come in bursts and that rhythm sells authenticity.

## Install

- Claude Code: `claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg`
- Codex: `npx skills add vargHQ/skills --skill varg-video-edit-library -a codex -y`
- Cursor: `npx skills add vargHQ/skills --skill varg-video-edit-library -a cursor -y`

## Example output

https://static.heygen.ai/hyperframes/templates/promoted/254fe4bd8a3f00c1e70ac5e38a1bce25060f7590/claude-exchange/preview.mp4 — Preview of the Claude Exchange component that builds this take (credit: Claude Exchange · hyperframes, https://hyperframes.heygen.com/catalog/blocks/claude-exchange)

## More AI agent flows skills

- [Prompt Typed, Then Generate](https://varg.ai/how-to/skills/prompt-typed-then-generate.md): A real task types into the prompt box, the Generate button is pressed, and the result starts to build.
- [Tool Calls as Status Rows](https://varg.ai/how-to/skills/tool-calls-status-rows.md): Each tool call appears as a row with an icon and label that goes spinner → check, building a visible plan.
- [Nodes Wire Up on a Canvas](https://varg.ai/how-to/skills/nodes-wire-up-canvas.md): Nodes pop onto a canvas and connectors draw between them as the camera pulls back to show the whole pipeline.
- [Beams Into the Agent Hub](https://varg.ai/how-to/skills/beams-into-agent-hub.md): Light pulses run along curved lines from tool icons into a central agent and back out to a result.
- [Thinking… Shimmer Label](https://varg.ai/how-to/skills/thinking-shimmer-label.md): A muted "Thinking…" label with a moving shine shows the model working between prompt and answer.
- [Parallel Agents Side by Side](https://varg.ai/how-to/skills/parallel-agents-lanes.md): Several agent cards work at once in lanes or an orbit, each with its own status, then merge into one result.

## Run the generation on varg

varg is one API key for the models these skills drive — Seedance 2.5, Kling 4, Veo 3.1, Wan 3, Nano Banana 2.1, GPT Image 2.5, ElevenLabs v4 and 200+ more — billed in credits, with a cost estimate before anything runs.

- Claude Code: `claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg` (skill + hosted MCP)
- Codex, Cursor, Gemini CLI, OpenClaw: `npx skills add vargHQ/skills --skill varg-ai -a <agent> -y`
- Grok and any agent that reads SKILL.md: `npx skills add vargHQ/skills --skill varg-ai`, or point it at the skill's SKILL.md
- Claude.ai, ChatGPT, any MCP client: connect `https://mcp.varg.ai/mcp`
- API key: https://varg.ai/dashboard/api · Model specs: https://varg.ai/models/llms.txt