varg.ai-sdk 0.1.0 → 0.4.0-alpha.1

package/action/video/SKILL.md

@@ -1,116 +0,0 @@
----
-name: video-generation
-description: generate videos from images or text prompts using fal.ai. use when user wants to animate images, create videos from text, or needs ai video generation with 5-10 second clips.
-allowed-tools: Read, Bash
----
-
-# video generation
-
-generate ai videos from images or text using fal.ai with automatic s3 upload support.
-
-## capabilities
-
-- **image-to-video**: animate static images with motion prompts
-- **text-to-video**: generate videos directly from text descriptions
-- supports 5 or 10 second duration
-- automatic s3 upload
-
-## usage
-
-### generate from image
-```bash
-bun run service/video.ts from_image <prompt> <imageUrl> [duration] [upload]
-```
-
-**parameters:**
-- `prompt` (required): motion description (e.g., "camera pan left")
-- `imageUrl` (required): url of the source image
-- `duration` (optional): 5 or 10 seconds (default: 5)
-- `upload` (optional): "true" to upload to s3
-
-**example:**
-```bash
-bun run service/video.ts from_image "person talking naturally" https://example.com/headshot.jpg 5 true
-```
-
-### generate from text
-```bash
-bun run service/video.ts from_text <prompt> [duration] [upload]
-```
-
-**parameters:**
-- `prompt` (required): video scene description
-- `duration` (optional): 5 or 10 seconds (default: 5)
-- `upload` (optional): "true" to upload to s3
-
-**example:**
-```bash
-bun run service/video.ts from_text "waves crashing on beach at sunset" 10 true
-```
-
-## as library
-
-```typescript
-import { generateVideoFromImage, generateVideoFromText } from "./service/video"
-
-// animate an image
-const videoResult = await generateVideoFromImage(
-  "camera zoom in slowly",
-  "https://example.com/portrait.jpg",
-  { duration: 5, upload: true }
-)
-console.log(videoResult.videoUrl)
-console.log(videoResult.uploaded) // s3 url if upload=true
-
-// generate from text
-const textVideo = await generateVideoFromText(
-  "forest path with sunlight filtering through trees",
-  { duration: 10, upload: true }
-)
-```
-
-## output
-
-returns `VideoGenerationResult`:
-```typescript
-{
-  videoUrl: string,      // direct video url
-  duration?: number,     // actual video duration
-  uploaded?: string      // s3 url if upload requested
-}
-```
-
-## when to use
-
-use this skill when:
-- animating character headshots or portraits
-- creating motion from static images
-- generating video clips from text descriptions
-- preparing videos for lipsync or editing pipeline
-- need short form video content (5-10s)
-
-## tips
-
-**for character animation:**
-- use subtle prompts like "person talking naturally" or "slight head movement"
-- keep duration at 5 seconds for character shots
-- combine with lipsync for talking videos
-
-**for scene generation:**
-- be descriptive about camera movement and scene dynamics
-- 10 seconds works better for landscape/scene videos
-
-## environment variables
-
-required:
-- `FAL_API_KEY` - for fal video generation
-
-optional (for s3 upload):
-- `CLOUDFLARE_R2_API_URL`
-- `CLOUDFLARE_ACCESS_KEY_ID`
-- `CLOUDFLARE_ACCESS_SECRET`
-- `CLOUDFLARE_R2_BUCKET`
-
-## generation time
-
-expect 2-3 minutes per video clip

package/action/video/index.ts

@@ -1,135 +0,0 @@
-#!/usr/bin/env bun
-/**
- * video generation service combining fal and higgsfield
- * usage: bun run service/video.ts <command> <args>
- */
-
-import { imageToVideo, textToVideo } from "../../lib/fal";
-import { uploadFromUrl } from "../../utilities/s3";
-
-export interface VideoGenerationResult {
-  videoUrl: string;
-  duration?: number;
-  uploaded?: string;
-}
-
-export async function generateVideoFromImage(
-  prompt: string,
-  imageUrl: string,
-  options: { duration?: 5 | 10; upload?: boolean } = {},
-): Promise<VideoGenerationResult> {
-  console.log("[service/video] generating video from image");
-
-  const result = await imageToVideo({
-    prompt,
-    imageUrl,
-    duration: options.duration,
-  });
-
-  const videoUrl = result.data?.video?.url;
-  if (!videoUrl) {
-    throw new Error("no video url in result");
-  }
-
-  let uploaded: string | undefined;
-  if (options.upload) {
-    const timestamp = Date.now();
-    const objectKey = `videos/generated/${timestamp}.mp4`;
-    uploaded = await uploadFromUrl(videoUrl, objectKey);
-    console.log(`[service/video] uploaded to ${uploaded}`);
-  }
-
-  return {
-    videoUrl,
-    duration: result.data?.duration,
-    uploaded,
-  };
-}
-
-export async function generateVideoFromText(
-  prompt: string,
-  options: { duration?: 5 | 10; upload?: boolean } = {},
-): Promise<VideoGenerationResult> {
-  console.log("[service/video] generating video from text");
-
-  const result = await textToVideo({
-    prompt,
-    duration: options.duration,
-  });
-
-  const videoUrl = result.data?.video?.url;
-  if (!videoUrl) {
-    throw new Error("no video url in result");
-  }
-
-  let uploaded: string | undefined;
-  if (options.upload) {
-    const timestamp = Date.now();
-    const objectKey = `videos/generated/${timestamp}.mp4`;
-    uploaded = await uploadFromUrl(videoUrl, objectKey);
-    console.log(`[service/video] uploaded to ${uploaded}`);
-  }
-
-  return {
-    videoUrl,
-    duration: result.data?.duration,
-    uploaded,
-  };
-}
-
-// cli runner
-if (import.meta.main) {
-  const [command, ...args] = process.argv.slice(2);
-
-  switch (command) {
-    case "from_image": {
-      if (!args[0] || !args[1]) {
-        console.log(`
-usage:
-  bun run service/video.ts from_image <prompt> <imageUrl> [duration] [upload]
-        `);
-        process.exit(1);
-      }
-      const duration = args[2];
-      if (duration && duration !== "5" && duration !== "10") {
-        console.error("duration must be 5 or 10");
-        process.exit(1);
-      }
-      const imgResult = await generateVideoFromImage(args[0], args[1], {
-        duration: duration === "10" ? 10 : 5,
-        upload: args[3] === "true",
-      });
-      console.log(JSON.stringify(imgResult, null, 2));
-      break;
-    }
-
-    case "from_text": {
-      if (!args[0]) {
-        console.log(`
-usage:
-  bun run service/video.ts from_text <prompt> [duration] [upload]
-        `);
-        process.exit(1);
-      }
-      const duration = args[1];
-      if (duration && duration !== "5" && duration !== "10") {
-        console.error("duration must be 5 or 10");
-        process.exit(1);
-      }
-      const txtResult = await generateVideoFromText(args[0], {
-        duration: duration === "10" ? 10 : 5,
-        upload: args[2] === "true",
-      });
-      console.log(JSON.stringify(txtResult, null, 2));
-      break;
-    }
-
-    default:
-      console.log(`
-usage:
-  bun run service/video.ts from_image <prompt> <imageUrl> [duration] [upload]
-  bun run service/video.ts from_text <prompt> [duration] [upload]
-      `);
-      process.exit(1);
-  }
-}

package/action/voice/SKILL.md

@@ -1,125 +0,0 @@
----
-name: voice-synthesis
-description: generate realistic text-to-speech audio using elevenlabs with multiple voice options. use when user needs voiceovers, narration, character voices, or audio for lipsync videos.
-allowed-tools: Read, Bash
----
-
-# voice synthesis
-
-generate high-quality text-to-speech audio with elevenlabs.
-
-## available voices
-
-- **rachel** - clear, professional female voice
-- **domi** - warm, friendly female voice
-- **bella** - energetic female voice
-- **antoni** - friendly male voice
-- **elli** - young, clear female voice
-- **josh** - deep, clear male voice
-- **arnold** - strong, authoritative male voice
-- **adam** - natural, conversational male voice
-- **sam** - raspy, character male voice
-
-## usage
-
-### generate voice
-```bash
-bun run service/voice.ts generate <text> [voice] [provider] [upload]
-```
-
-**parameters:**
-- `text` (required): text to convert to speech
-- `voice` (optional): voice name (default: rachel)
-- `provider` (optional): elevenlabs (default)
-- `upload` (optional): "true" to upload to s3
-
-**example:**
-```bash
-bun run service/voice.ts generate "hello world, this is my voice" rachel elevenlabs true
-```
-
-### shorthand for elevenlabs
-```bash
-bun run service/voice.ts elevenlabs <text> [voice] [upload]
-```
-
-**example:**
-```bash
-bun run service/voice.ts elevenlabs "welcome to our video" josh true
-```
-
-## as library
-
-```typescript
-import { generateVoice } from "./service/voice"
-
-const result = await generateVoice({
-  text: "hello world",
-  voice: "rachel",
-  provider: "elevenlabs",
-  upload: true,
-  outputPath: "media/voiceover.mp3"
-})
-
-console.log(result.provider)
-console.log(result.voiceId)
-console.log(result.uploadUrl)
-```
-
-## output
-
-returns `VoiceResult`:
-```typescript
-{
-  audio: Buffer,         // raw audio buffer
-  provider: string,      // "elevenlabs"
-  voiceId: string,       // actual voice id used
-  uploadUrl?: string     // s3 url if upload requested
-}
-```
-
-saves audio file to `media/voice-{timestamp}.mp3`
-
-## when to use
-
-use this skill when:
-- creating voiceovers for videos
-- generating narration or character dialogue
-- preparing audio for lipsync videos
-- need text-to-speech for talking character pipeline
-- testing different voice options
-
-## tips
-
-**voice selection:**
-- use **rachel** or **josh** for professional narration
-- use **bella** or **antoni** for friendly, casual content
-- use **arnold** for authoritative or dramatic content
-- use **sam** for character or stylized voices
-
-**text formatting:**
-- add punctuation for natural pauses
-- use shorter sentences for clearer speech
-- spell out numbers and abbreviations
-
-## integration with other services
-
-perfect companion for:
-- **lipsync service** - sync generated voice with video
-- **video generation** - create talking character videos
-- **captions service** - auto-generate subtitles from voiceover
-
-## environment variables
-
-required:
-- `ELEVENLABS_API_KEY` - for voice generation
-
-optional (for s3 upload):
-- `CLOUDFLARE_R2_API_URL`
-- `CLOUDFLARE_ACCESS_KEY_ID`
-- `CLOUDFLARE_ACCESS_SECRET`
-- `CLOUDFLARE_R2_BUCKET`
-
-## generation time
-
-expect 5-15 seconds depending on text length

package/action/voice/index.ts

@@ -1,201 +0,0 @@
-#!/usr/bin/env bun
-
-/**
- * voice service - high-level voice generation combining multiple providers
- * supports elevenlabs and future providers
- */
-
-import { textToSpeech, VOICES } from "../../lib/elevenlabs";
-import { uploadFile } from "../../utilities/s3";
-
-// types
-export interface GenerateVoiceOptions {
-  text: string;
-  voice?: string;
-  provider?: "elevenlabs";
-  upload?: boolean;
-  outputPath?: string;
-}
-
-export interface VoiceResult {
-  audio: Buffer;
-  provider: string;
-  voiceId: string;
-  uploadUrl?: string;
-}
-
-// core functions
-export async function generateVoice(
-  options: GenerateVoiceOptions,
-): Promise<VoiceResult> {
-  const {
-    text,
-    voice = "rachel",
-    provider = "elevenlabs",
-    upload = false,
-    outputPath,
-  } = options;
-
-  if (!text) {
-    throw new Error("text is required");
-  }
-
-  console.log(`[voice] generating with ${provider} (${voice})...`);
-
-  let audio: Buffer;
-  let voiceId: string;
-
-  switch (provider) {
-    case "elevenlabs": {
-      // map friendly names to voice ids
-      const voiceMap: Record<string, string> = {
-        rachel: VOICES.RACHEL,
-        domi: VOICES.DOMI,
-        bella: VOICES.BELLA,
-        antoni: VOICES.ANTONI,
-        elli: VOICES.ELLI,
-        josh: VOICES.JOSH,
-        arnold: VOICES.ARNOLD,
-        adam: VOICES.ADAM,
-        sam: VOICES.SAM,
-      };
-
-      voiceId = voiceMap[voice.toLowerCase()] || voice;
-
-      audio = await textToSpeech({
-        text,
-        voiceId,
-        outputPath,
-      });
-      break;
-    }
-
-    default:
-      throw new Error(`unsupported provider: ${provider}`);
-  }
-
-  const result: VoiceResult = {
-    audio,
-    provider,
-    voiceId,
-  };
-
-  // upload to s3 if requested
-  if (upload && outputPath) {
-    const objectKey = `voice/${Date.now()}-${voice}.mp3`;
-    const uploadUrl = await uploadFile(outputPath, objectKey);
-    result.uploadUrl = uploadUrl;
-    console.log(`[voice] uploaded to ${uploadUrl}`);
-  }
-
-  return result;
-}
-
-// cli
-async function cli() {
-  const args = process.argv.slice(2);
-  const command = args[0];
-
-  if (!command || command === "help") {
-    console.log(`
-usage:
-  bun run service/voice.ts <command> [args]
-
-commands:
-  generate <text> [voice] [provider] [upload]    generate voice from text
-  elevenlabs <text> [voice] [upload]             generate with elevenlabs
-  help                                           show this help
-
-examples:
-  bun run service/voice.ts generate "hello world" rachel elevenlabs false
-  bun run service/voice.ts elevenlabs "hello world" josh true
-  bun run service/voice.ts generate "welcome to ai" bella
-
-available voices:
-  rachel, domi, bella, antoni, elli, josh, arnold, adam, sam
-
-providers:
-  elevenlabs (default)
-
-environment:
-  ELEVENLABS_API_KEY - required for elevenlabs
-  CLOUDFLARE_* - required for upload
-    `);
-    process.exit(0);
-  }
-
-  try {
-    switch (command) {
-      case "generate": {
-        const text = args[1];
-        const voice = args[2];
-        const provider = (args[3] || "elevenlabs") as "elevenlabs";
-        const upload = args[4] === "true";
-
-        if (!text) {
-          throw new Error("text is required");
-        }
-
-        const outputPath = `media/voice-${Date.now()}.mp3`;
-
-        const result = await generateVoice({
-          text,
-          voice,
-          provider,
-          upload,
-          outputPath,
-        });
-
-        console.log(`[voice] result:`, {
-          provider: result.provider,
-          voiceId: result.voiceId,
-          audioSize: result.audio.length,
-          outputPath,
-          uploadUrl: result.uploadUrl,
-        });
-        break;
-      }
-
-      case "elevenlabs": {
-        const text = args[1];
-        const voice = args[2];
-        const upload = args[3] === "true";
-
-        if (!text) {
-          throw new Error("text is required");
-        }
-
-        const outputPath = `media/voice-${Date.now()}.mp3`;
-
-        const result = await generateVoice({
-          text,
-          voice,
-          provider: "elevenlabs",
-          upload,
-          outputPath,
-        });
-
-        console.log(`[voice] result:`, {
-          provider: result.provider,
-          voiceId: result.voiceId,
-          audioSize: result.audio.length,
-          outputPath,
-          uploadUrl: result.uploadUrl,
-        });
-        break;
-      }
-
-      default:
-        console.error(`unknown command: ${command}`);
-        console.log(`run 'bun run service/voice.ts help' for usage`);
-        process.exit(1);
-    }
-  } catch (error) {
-    console.error(`[voice] error:`, error);
-    process.exit(1);
-  }
-}
-
-if (import.meta.main) {
-  cli();
-}

package/index.ts

@@ -1,38 +0,0 @@
-/**
- * varg.ai sdk
- * video generation and editing tools
- */
-
-// re-export external clients
-export { fal } from "@ai-sdk/fal";
-export { replicate } from "@ai-sdk/replicate";
-export { fal as falClient } from "@fal-ai/client";
-export { HiggsfieldClient } from "@higgsfield/client";
-// lib exports - ai-sdk/fal (provider)
-export * as aiSdkFal from "./lib/ai-sdk/fal";
-// lib exports - ai-sdk/replicate (provider)
-export * as aiSdkReplicate from "./lib/ai-sdk/replicate";
-// lib exports - elevenlabs
-export * from "./lib/elevenlabs";
-// lib exports - fal (client)
-export * from "./lib/fal";
-// lib exports - ffmpeg
-export * from "./lib/ffmpeg";
-// lib exports - fireworks
-export * from "./lib/fireworks";
-// lib exports - groq
-export * from "./lib/groq";
-// lib exports - higgsfield
-export * from "./lib/higgsfield";
-// lib exports - replicate
-export * from "./lib/replicate";
-// service exports
-export * from "./service/captions";
-export * from "./service/edit";
-export * from "./service/image";
-export * from "./service/sync";
-export * from "./service/transcribe";
-export * from "./service/video";
-export * from "./service/voice";
-// utilities exports
-export * from "./utilities/s3";