Skip to content

API.md

Latest commit

 

History

History
237 lines (170 loc) · 5.68 KB

API.md

File metadata and controls

237 lines (170 loc) · 5.68 KB
Raw

API Reference

Base URL: https://api.llamagen.ai/v1

This package covers LlamaGen's Comic API and Animation API. Type definitions live in ../src/api-types.ts.

Client

import { LlamaGenClient } from 'comic';

const llamagen = new LlamaGenClient({
  apiKey: process.env.LLAMAGEN_API_KEY!,
  timeoutMs: 30000,
  maxRetries: 2,
  retryDelayMs: 500
});

Comic API

llamagen.comic.create(params)

Creates a comic generation through POST /comics/generations.

const created = await llamagen.comic.create({
  prompt: 'A 4-panel comic about Leo finding a glowing key in a quiet library.',
  style: 'manga',
  fixPanelNum: 4
});

Supported input families:

  • Story input: prompt, promptUrl, style, preset, size.
  • Layout: fixPanelNum for one page, or pagination.totalPages plus pagination.panelsPerPage for multi-page mode.
  • Continuity references: comicRoles, comicLocations, images, attachments.
  • Output controls: language, upscale.

The SDK validates that prompt is non-empty, supported size values are used, and fixPanelNum is not mixed with pagination.

llamagen.comic.get(id, options?)

Gets a generation through GET /comics/generations/{generationId}.

const generation = await llamagen.comic.get('gen_123');
const singlePanel = await llamagen.comic.get('gen_123', { page: 0, panel: 2 });

page and panel are zero-based.

llamagen.comic.continueWrite(id, params)

Extends an existing story through PATCH /comics/generations/{generationId} with action: "continueWrite".

await llamagen.comic.continueWrite('gen_123', {
  prompt: 'The two friends walk deeper into the neon city and find a hidden arcade.',
  pagination: { totalPages: 1, panelsPerPage: 4 },
  attachments: [{ type: 'image', url: 'https://example.com/reference.png' }]
});

llamagen.comic.updatePanel(id, params)

Regenerates one panel through PATCH /comics/generations/{generationId} with action: "regeneratePanel".

await llamagen.comic.updatePanel('gen_123', {
  page: 0,
  panel: 2,
  panelPrompt: 'Make Leo look more hopeful and keep the same orange wizard robe.'
});

Accepted aliases include panelIndex, panel_index, pageIndex, page_index, prompt, panel_prompt, and images_url.

Usage, Uploads, and Polling

const usage = await llamagen.comic.usage();
const upload = await llamagen.comic.upload(file, 'reference.png');
const done = await llamagen.comic.waitForCompletion('gen_123');
const createdAndDone = await llamagen.comic.createAndWait({ prompt: 'A detective fox in Tokyo.' });

Batch helpers:

  • llamagen.comic.createBatch(paramsList, { concurrency, stopOnError })
  • llamagen.comic.waitForMany(ids, { concurrency, intervalMs, timeoutMs })

Animation API

llamagen.animation.create(params)

Creates a video artwork through POST /artworks/generations.

const created = await llamagen.animation.create({
  prompt: 'A heroic fox detective walks through neon rain, cinematic camera move.',
  videoOptions: {
    duration: 5,
    resolution: '720p',
    aspect_ratio: '16:9',
    image: 'https://example.com/first-frame.png',
    last_frame_image: 'https://example.com/last-frame.png',
    reference_images: ['https://example.com/character.png']
  }
});

Animation methods:

  • llamagen.animation.create(params)
  • llamagen.animation.get(id)
  • llamagen.animation.waitForCompletion(id, options?)
  • llamagen.animation.createAndWait(params, options?)

Use this namespace for text-to-video, image-to-video, storyboard-to-video, and reference-driven animation workflows.

Webhooks

import { constructWebhookEvent } from 'comic';

const event = constructWebhookEvent(rawBody, headers, process.env.LLAMAGEN_WEBHOOK_SECRET!);

Supported event names include:

  • comic.generation.created
  • comic.generation.updated
  • comic.generation.completed
  • comic.generation.failed

Headers verified by the SDK:

  • X-Llama-Webhook-Id
  • X-Llama-Webhook-Timestamp
  • X-Llama-Webhook-Signature
  • X-Llama-Webhook-Request-Id

Direct HTTP

Comic endpoints:

  • POST /comics/generations
  • GET /comics/generations/{generationId}
  • PATCH /comics/generations/{generationId}
  • GET /comics/usage
  • POST /comics/upload

Animation endpoints:

  • POST /artworks/generations
  • GET /artworks/generations/{id}
curl -X POST https://api.llamagen.ai/v1/comics/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A fox detective in Tokyo.",
    "style": "manga",
    "fixPanelNum": 4
  }'

MCP

Remote MCP endpoint:

https://llamagen.ai/api/mcp

Available MCP tools:

  • create_comic_generation
  • get_comic_generation_status
  • get_api_usage

Framework Integrations

MCP Usage

Use MCP Streamable HTTP endpoint:

https://llamagen.ai/api/mcp

Authorization:

Authorization: Bearer YOUR_API_KEY

Available tools:

  • create_comic_generation
  • get_comic_generation_status
  • get_api_usage

Client config example:

{
  "mcpServers": {
    "llamagen": {
      "url": "https://llamagen.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Cursor setup:

  1. Go to Cursor MCP settings.
  2. Add a new Streamable HTTP server.
  3. Set URL to https://llamagen.ai/api/mcp.
  4. Add header Authorization: Bearer YOUR_API_KEY.
  5. Confirm tool list is visible.

OAuth metadata endpoint:

https://llamagen.ai/.well-known/oauth-protected-resource