# Cliphi for developers

Turn your long video into ready-to-post vertical clips from your agent, app, or workflow. One API, previews first, renders billed.

Cliphi's Agent API takes your video's URL, finds the strongest moments, and
returns finished vertical clips with captions and branding. Free previews come
first; rendering is the billed choice.

## Pick your surface

- **REST**: `https://www.cliphi.com/api/v1`: submit, status, render, plus
  a keyless demo. Start with the [quickstart](/docs/quickstart).
- **MCP**: `https://www.cliphi.com/api/mcp`: the same API as MCP tools for
  Claude connectors, Cursor, Gemini, n8n and any MCP client. See [MCP
  server](/docs/mcp).
- **ChatGPT Actions**: import the [OpenAPI schema](https://www.cliphi.com/cliphi-actions.json)
  into a custom GPT. See [ChatGPT Actions](/docs/chatgpt-actions).
- **Agent skill**: a ready-made SKILL.md for OpenClaw and skill-compatible
  agents, guardrails baked in; published on the skill registries at launch.

## The full walkthrough

The [Agent API walkthrough](/docs/agents) carries every request and response
shape end to end, plus a keyless live demo you can curl right now:

```bash
curl https://www.cliphi.com/api/v1/demo
```

## The rules every integration should keep

1. Previews first: show the free `preview_page_url` links and get an explicit
   pick before rendering.
2. Renders cost credits and state their price in the response
   (`estimated_credits`). Never render unprompted.
3. Relay `message` fields verbatim; branch on the `status` FIELD, never assume
   an HTTP 200 means success.

Not building anything? [Use Cliphi from ChatGPT or Claude instead](/agents).