vargai 0.4.0-alpha89 → 0.4.0-alpha90

package/package.json

@@ -71,7 +71,7 @@
     "zod": "^4.2.1"
   },
   "sideEffects": false,
-  "version": "0.4.0-alpha89",
+  "version": "0.4.0-alpha90",
   "exports": {
     ".": "./src/index.ts",
     "./ai": "./src/ai-sdk/index.ts",

package/src/definitions/skills/index.ts

@@ -0,0 +1,12 @@
+/**
+ * Skill definitions index
+ */
+
+export { definition as talkingCharacter } from "./talking-character";
+export { definition as textToTiktok } from "./text-to-tiktok";
+
+// All skill definitions for auto-loading
+import { definition as talkingCharacterDefinition } from "./talking-character";
+import { definition as textToTiktokDefinition } from "./text-to-tiktok";
+
+export const allSkills = [talkingCharacterDefinition, textToTiktokDefinition];

package/src/definitions/skills/talking-character.ts

@@ -0,0 +1,87 @@
+/**
+ * Talking Character Skill
+ * Create a talking character video with lipsync and captions
+ */
+
+import { z } from "zod";
+import {
+  captionStyleSchema,
+  simpleVoiceSchema,
+  videoDurationSchema,
+} from "../../core/schema/shared";
+import type { SkillDefinition, ZodSchema } from "../../core/schema/types";
+
+// Input schema with Zod
+const talkingCharacterInputSchema = z.object({
+  text: z.string().describe("Script/text for the character to say"),
+  characterPrompt: z
+    .string()
+    .default("professional headshot of a friendly person, studio lighting")
+    .describe("Prompt to generate the character"),
+  voice: simpleVoiceSchema.default("sam").describe("Voice to use for speech"),
+  duration: videoDurationSchema.default(5).describe("Video duration"),
+  style: captionStyleSchema.default("tiktok").describe("Caption style"),
+});
+
+// Output schema with Zod
+const talkingCharacterOutputSchema = z.object({
+  videoUrl: z.string(),
+  characterImageUrl: z.string().optional(),
+  audioPath: z.string().optional(),
+});
+
+// Schema object for the definition
+const schema: ZodSchema<
+  typeof talkingCharacterInputSchema,
+  typeof talkingCharacterOutputSchema
+> = {
+  input: talkingCharacterInputSchema,
+  output: talkingCharacterOutputSchema,
+};
+
+export const definition: SkillDefinition<typeof schema> = {
+  type: "skill",
+  name: "talking-character",
+  description: "Create a talking character video with lipsync and captions",
+  schema,
+  steps: [
+    {
+      name: "generate-character",
+      run: "image",
+      inputs: {
+        prompt: "$inputs.characterPrompt",
+        provider: "higgsfield",
+      },
+    },
+    {
+      name: "generate-voice",
+      run: "voice",
+      inputs: {
+        text: "$inputs.text",
+        voice: "$inputs.voice",
+        output: "output/voiceover.mp3",
+      },
+    },
+    {
+      name: "animate-character",
+      run: "sync",
+      inputs: {
+        image: "$results.generate-character.imageUrl",
+        audio: "output/voiceover.mp3",
+        prompt: "person talking naturally, professional demeanor",
+        duration: "$inputs.duration",
+      },
+    },
+    {
+      name: "add-captions",
+      run: "captions",
+      inputs: {
+        video: "$results.animate-character.videoUrl",
+        output: "output/final.mp4",
+        style: "$inputs.style",
+      },
+    },
+  ],
+};
+
+export default definition;

package/src/definitions/skills/text-to-tiktok.ts

@@ -0,0 +1,97 @@
+/**
+ * Text to TikTok Skill
+ * Turn text into a TikTok with AI-generated looping background and voiceover
+ */
+
+import { z } from "zod";
+import {
+  captionStyleSchema,
+  simpleVoiceSchema,
+} from "../../core/schema/shared";
+import type { SkillDefinition, ZodSchema } from "../../core/schema/types";
+
+// Input schema with Zod
+const textToTiktokInputSchema = z.object({
+  text: z.string().describe("Text content to convert to video"),
+  voice: simpleVoiceSchema.default("sam").describe("Voice for narration"),
+  backgroundPrompt: z
+    .string()
+    .default(
+      "POV from inside moving car driving through rainy city at night, motion blur on streetlights, cinematic",
+    )
+    .describe("Prompt for background video"),
+  captionStyle: captionStyleSchema.default("tiktok").describe("Caption style"),
+});
+
+// Output schema with Zod
+const textToTiktokOutputSchema = z.object({
+  videoUrl: z.string(),
+  voiceoverPath: z.string().optional(),
+  captionsPath: z.string().optional(),
+  backgroundVideoUrl: z.string().optional(),
+});
+
+// Schema object for the definition
+const schema: ZodSchema<
+  typeof textToTiktokInputSchema,
+  typeof textToTiktokOutputSchema
+> = {
+  input: textToTiktokInputSchema,
+  output: textToTiktokOutputSchema,
+};
+
+export const definition: SkillDefinition<typeof schema> = {
+  type: "skill",
+  name: "text-to-tiktok",
+  description: "Turn text into a TikTok with looping background and voiceover",
+  schema,
+  steps: [
+    {
+      name: "generate-voiceover",
+      run: "voice",
+      inputs: {
+        text: "$inputs.text",
+        voice: "$inputs.voice",
+        output: "output/voiceover.mp3",
+      },
+    },
+    {
+      name: "transcribe",
+      run: "transcribe",
+      inputs: {
+        audio: "output/voiceover.mp3",
+        provider: "fireworks",
+        output: "output/captions.srt",
+      },
+    },
+    {
+      name: "generate-background-frame",
+      run: "image",
+      inputs: {
+        prompt: "$inputs.backgroundPrompt",
+        size: "portrait_16_9",
+      },
+    },
+    {
+      name: "generate-background-video",
+      run: "video",
+      inputs: {
+        prompt: "$inputs.backgroundPrompt",
+        image: "$results.generate-background-frame.imageUrl",
+        duration: 10,
+      },
+    },
+    {
+      name: "add-captions",
+      run: "captions",
+      inputs: {
+        video: "$results.generate-background-video.videoUrl",
+        output: "output/final.mp4",
+        srt: "output/captions.srt",
+        style: "$inputs.captionStyle",
+      },
+    },
+  ],
+};
+
+export default definition;