vidspotai-shared 1.0.67 → 1.0.69-dev.0

package/lib/schemas/brief.schema.d.ts

@@ -0,0 +1,263 @@
+import { z } from "zod";
+/**
+ * VideoBrief — the structured spec the agent fills from a user's natural-language
+ * request. It is the single source of truth for downstream tools (planner, executor,
+ * critic). Agents are stateless across turns; the Brief carries state.
+ *
+ * Two-stage intent capture produces this:
+ *   1. Implicit inference pass: LLM fills confident defaults from the prompt + niche.
+ *   2. Targeted gap pass: chip-style clarifying questions for low-confidence axes only.
+ *
+ * Fields are mostly optional because partial briefs are valid early in the conversation;
+ * the planner re-validates that required-for-planning fields are present at plan time.
+ */
+export declare const VideoNicheSchema: z.ZodEnum<{
+    other: "other";
+    news: "news";
+    "ugc-ad": "ugc-ad";
+    "faceless-yt": "faceless-yt";
+    "product-demo": "product-demo";
+    "talking-head": "talking-head";
+    "social-recap": "social-recap";
+    "real-estate": "real-estate";
+    recipe: "recipe";
+    travel: "travel";
+    fitness: "fitness";
+    "podcast-clip": "podcast-clip";
+    education: "education";
+}>;
+export declare const ToneSchema: z.ZodEnum<{
+    professional: "professional";
+    energetic: "energetic";
+    calm: "calm";
+    cinematic: "cinematic";
+    casual: "casual";
+    playful: "playful";
+    edgy: "edgy";
+    warm: "warm";
+    informational: "informational";
+}>;
+export declare const HookStyleSchema: z.ZodEnum<{
+    question: "question";
+    "bold-claim": "bold-claim";
+    "pattern-interrupt": "pattern-interrupt";
+    story: "story";
+    stat: "stat";
+    demo: "demo";
+    "voiceover-only": "voiceover-only";
+}>;
+export declare const PacingSchema: z.ZodEnum<{
+    medium: "medium";
+    fast: "fast";
+    slow: "slow";
+    "very-fast": "very-fast";
+}>;
+export declare const VoiceStyleSchema: z.ZodEnum<{
+    energetic: "energetic";
+    calm: "calm";
+    casual: "casual";
+    warm: "warm";
+    neutral: "neutral";
+    excited: "excited";
+    happy: "happy";
+    serious: "serious";
+    dramatic: "dramatic";
+    whisper: "whisper";
+}>;
+export declare const VoiceProfileSchema: z.ZodObject<{
+    gender: z.ZodOptional<z.ZodEnum<{
+        male: "male";
+        female: "female";
+        neutral: "neutral";
+    }>>;
+    ageRange: z.ZodOptional<z.ZodEnum<{
+        child: "child";
+        teen: "teen";
+        "young-adult": "young-adult";
+        adult: "adult";
+        senior: "senior";
+    }>>;
+    accent: z.ZodOptional<z.ZodString>;
+    language: z.ZodDefault<z.ZodString>;
+    voiceId: z.ZodOptional<z.ZodString>;
+    style: z.ZodOptional<z.ZodEnum<{
+        energetic: "energetic";
+        calm: "calm";
+        casual: "casual";
+        warm: "warm";
+        neutral: "neutral";
+        excited: "excited";
+        happy: "happy";
+        serious: "serious";
+        dramatic: "dramatic";
+        whisper: "whisper";
+    }>>;
+}, z.core.$strip>;
+export declare const MusicProfileSchema: z.ZodObject<{
+    moods: z.ZodDefault<z.ZodArray<z.ZodEnum<{
+        energetic: "energetic";
+        calm: "calm";
+        cinematic: "cinematic";
+        playful: "playful";
+        uplifting: "uplifting";
+        tense: "tense";
+        dreamy: "dreamy";
+        dark: "dark";
+    }>>>;
+    bpmMin: z.ZodOptional<z.ZodNumber>;
+    bpmMax: z.ZodOptional<z.ZodNumber>;
+}, z.core.$strip>;
+export declare const ClaimsSchema: z.ZodObject<{
+    allow: z.ZodDefault<z.ZodArray<z.ZodString>>;
+    block: z.ZodDefault<z.ZodArray<z.ZodString>>;
+}, z.core.$strip>;
+export declare const ReferenceLinkSchema: z.ZodObject<{
+    url: z.ZodString;
+    analyzed: z.ZodOptional<z.ZodObject<{
+        durationMs: z.ZodNumber;
+        sceneCount: z.ZodNumber;
+        hookMs: z.ZodNumber;
+        cutsPerSecond: z.ZodNumber;
+        bpm: z.ZodOptional<z.ZodNumber>;
+        transcriptExcerpt: z.ZodOptional<z.ZodString>;
+    }, z.core.$strip>>;
+}, z.core.$strip>;
+export declare const VideoBriefSchema: z.ZodObject<{
+    rawPrompt: z.ZodString;
+    goal: z.ZodOptional<z.ZodString>;
+    niche: z.ZodOptional<z.ZodEnum<{
+        other: "other";
+        news: "news";
+        "ugc-ad": "ugc-ad";
+        "faceless-yt": "faceless-yt";
+        "product-demo": "product-demo";
+        "talking-head": "talking-head";
+        "social-recap": "social-recap";
+        "real-estate": "real-estate";
+        recipe: "recipe";
+        travel: "travel";
+        fitness: "fitness";
+        "podcast-clip": "podcast-clip";
+        education: "education";
+    }>>;
+    audience: z.ZodOptional<z.ZodString>;
+    tone: z.ZodOptional<z.ZodEnum<{
+        professional: "professional";
+        energetic: "energetic";
+        calm: "calm";
+        cinematic: "cinematic";
+        casual: "casual";
+        playful: "playful";
+        edgy: "edgy";
+        warm: "warm";
+        informational: "informational";
+    }>>;
+    hookStyle: z.ZodOptional<z.ZodEnum<{
+        question: "question";
+        "bold-claim": "bold-claim";
+        "pattern-interrupt": "pattern-interrupt";
+        story: "story";
+        stat: "stat";
+        demo: "demo";
+        "voiceover-only": "voiceover-only";
+    }>>;
+    pacing: z.ZodOptional<z.ZodEnum<{
+        medium: "medium";
+        fast: "fast";
+        slow: "slow";
+        "very-fast": "very-fast";
+    }>>;
+    durationSec: z.ZodOptional<z.ZodNumber>;
+    aspect: z.ZodOptional<z.ZodEnum<{
+        "9:16": "9:16";
+        "1:1": "1:1";
+        "4:5": "4:5";
+        "16:9": "16:9";
+    }>>;
+    voice: z.ZodOptional<z.ZodObject<{
+        gender: z.ZodOptional<z.ZodEnum<{
+            male: "male";
+            female: "female";
+            neutral: "neutral";
+        }>>;
+        ageRange: z.ZodOptional<z.ZodEnum<{
+            child: "child";
+            teen: "teen";
+            "young-adult": "young-adult";
+            adult: "adult";
+            senior: "senior";
+        }>>;
+        accent: z.ZodOptional<z.ZodString>;
+        language: z.ZodDefault<z.ZodString>;
+        voiceId: z.ZodOptional<z.ZodString>;
+        style: z.ZodOptional<z.ZodEnum<{
+            energetic: "energetic";
+            calm: "calm";
+            casual: "casual";
+            warm: "warm";
+            neutral: "neutral";
+            excited: "excited";
+            happy: "happy";
+            serious: "serious";
+            dramatic: "dramatic";
+            whisper: "whisper";
+        }>>;
+    }, z.core.$strip>>;
+    music: z.ZodOptional<z.ZodObject<{
+        moods: z.ZodDefault<z.ZodArray<z.ZodEnum<{
+            energetic: "energetic";
+            calm: "calm";
+            cinematic: "cinematic";
+            playful: "playful";
+            uplifting: "uplifting";
+            tense: "tense";
+            dreamy: "dreamy";
+            dark: "dark";
+        }>>>;
+        bpmMin: z.ZodOptional<z.ZodNumber>;
+        bpmMax: z.ZodOptional<z.ZodNumber>;
+    }, z.core.$strip>>;
+    cta: z.ZodOptional<z.ZodString>;
+    claims: z.ZodDefault<z.ZodObject<{
+        allow: z.ZodDefault<z.ZodArray<z.ZodString>>;
+        block: z.ZodDefault<z.ZodArray<z.ZodString>>;
+    }, z.core.$strip>>;
+    references: z.ZodDefault<z.ZodArray<z.ZodObject<{
+        url: z.ZodString;
+        analyzed: z.ZodOptional<z.ZodObject<{
+            durationMs: z.ZodNumber;
+            sceneCount: z.ZodNumber;
+            hookMs: z.ZodNumber;
+            cutsPerSecond: z.ZodNumber;
+            bpm: z.ZodOptional<z.ZodNumber>;
+            transcriptExcerpt: z.ZodOptional<z.ZodString>;
+        }, z.core.$strip>>;
+    }, z.core.$strip>>>;
+    tierBudget: z.ZodDefault<z.ZodEnum<{
+        T0: "T0";
+        T1: "T1";
+        T2: "T2";
+        T3: "T3";
+    }>>;
+    videoStrategy: z.ZodOptional<z.ZodEnum<{
+        auto: "auto";
+        direct: "direct";
+        indirect: "indirect";
+    }>>;
+    consistencyTier: z.ZodOptional<z.ZodEnum<{
+        C0: "C0";
+        C1: "C1";
+        C2: "C2";
+    }>>;
+    confidence: z.ZodDefault<z.ZodRecord<z.ZodString, z.ZodNumber>>;
+}, z.core.$strip>;
+export type VideoBrief = z.infer<typeof VideoBriefSchema>;
+export type VoiceProfile = z.infer<typeof VoiceProfileSchema>;
+export type VoiceStyle = z.infer<typeof VoiceStyleSchema>;
+export type MusicProfile = z.infer<typeof MusicProfileSchema>;
+export type ReferenceLink = z.infer<typeof ReferenceLinkSchema>;
+export type VideoNiche = z.infer<typeof VideoNicheSchema>;
+export type Tone = z.infer<typeof ToneSchema>;
+export type HookStyle = z.infer<typeof HookStyleSchema>;
+export type Pacing = z.infer<typeof PacingSchema>;
+//# sourceMappingURL=brief.schema.d.ts.map

package/lib/schemas/brief.schema.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"brief.schema.d.ts","sourceRoot":"","sources":["../../src/schemas/brief.schema.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAGxB;;;;;;;;;;;GAWG;AAEH,eAAO,MAAM,gBAAgB;;;;;;;;;;;;;;EAc3B,CAAC;AAEH,eAAO,MAAM,UAAU;;;;;;;;;;EAUrB,CAAC;AAEH,eAAO,MAAM,eAAe;;;;;;;;EAQ1B,CAAC;AAEH,eAAO,MAAM,YAAY;;;;;EAAkD,CAAC;AAK5E,eAAO,MAAM,gBAAgB;;;;;;;;;;;EAW3B,CAAC;AAEH,eAAO,MAAM,kBAAkB;;;;;;;;;;;;;;;;;;;;;;;;;;;;iBAS7B,CAAC;AAEH,eAAO,MAAM,kBAAkB;;;;;;;;;;;;;iBAiB7B,CAAC;AAEH,eAAO,MAAM,YAAY;;;iBAKvB,CAAC;AAEH,eAAO,MAAM,mBAAmB;;;;;;;;;;iBAa9B,CAAC;AAEH,eAAO,MAAM,gBAAgB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;iBAoD3B,CAAC;AAEH,MAAM,MAAM,UAAU,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,gBAAgB,CAAC,CAAC;AAC1D,MAAM,MAAM,YAAY,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,kBAAkB,CAAC,CAAC;AAC9D,MAAM,MAAM,UAAU,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,gBAAgB,CAAC,CAAC;AAC1D,MAAM,MAAM,YAAY,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,kBAAkB,CAAC,CAAC;AAC9D,MAAM,MAAM,aAAa,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,mBAAmB,CAAC,CAAC;AAChE,MAAM,MAAM,UAAU,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,gBAAgB,CAAC,CAAC;AAC1D,MAAM,MAAM,IAAI,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,UAAU,CAAC,CAAC;AAC9C,MAAM,MAAM,SAAS,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,eAAe,CAAC,CAAC;AACxD,MAAM,MAAM,MAAM,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,YAAY,CAAC,CAAC"}

package/lib/schemas/brief.schema.js

@@ -0,0 +1,157 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.VideoBriefSchema = exports.ReferenceLinkSchema = exports.ClaimsSchema = exports.MusicProfileSchema = exports.VoiceProfileSchema = exports.VoiceStyleSchema = exports.PacingSchema = exports.HookStyleSchema = exports.ToneSchema = exports.VideoNicheSchema = void 0;
+const zod_1 = require("zod");
+const project_schema_1 = require("./project.schema");
+/**
+ * VideoBrief — the structured spec the agent fills from a user's natural-language
+ * request. It is the single source of truth for downstream tools (planner, executor,
+ * critic). Agents are stateless across turns; the Brief carries state.
+ *
+ * Two-stage intent capture produces this:
+ *   1. Implicit inference pass: LLM fills confident defaults from the prompt + niche.
+ *   2. Targeted gap pass: chip-style clarifying questions for low-confidence axes only.
+ *
+ * Fields are mostly optional because partial briefs are valid early in the conversation;
+ * the planner re-validates that required-for-planning fields are present at plan time.
+ */
+exports.VideoNicheSchema = zod_1.z.enum([
+    "ugc-ad",
+    "faceless-yt",
+    "product-demo",
+    "talking-head",
+    "social-recap",
+    "real-estate",
+    "recipe",
+    "travel",
+    "fitness",
+    "podcast-clip",
+    "education",
+    "news",
+    "other",
+]);
+exports.ToneSchema = zod_1.z.enum([
+    "energetic",
+    "calm",
+    "cinematic",
+    "casual",
+    "professional",
+    "playful",
+    "edgy",
+    "warm",
+    "informational",
+]);
+exports.HookStyleSchema = zod_1.z.enum([
+    "question",
+    "bold-claim",
+    "pattern-interrupt",
+    "story",
+    "stat",
+    "demo",
+    "voiceover-only",
+]);
+exports.PacingSchema = zod_1.z.enum(["slow", "medium", "fast", "very-fast"]);
+// AG-31: VoiceStyleSchema — drives ElevenLabs voice_settings (style + stability +
+// speed). Mapping is in executor.ts → resolveVoiceSettings(). Keep these as
+// adjectives the brief author can pick from the UI, not raw 0-1 numbers.
+exports.VoiceStyleSchema = zod_1.z.enum([
+    "neutral",
+    "calm",
+    "warm",
+    "casual",
+    "excited",
+    "energetic",
+    "happy",
+    "serious",
+    "dramatic",
+    "whisper",
+]);
+exports.VoiceProfileSchema = zod_1.z.object({
+    gender: zod_1.z.enum(["female", "male", "neutral"]).optional(),
+    ageRange: zod_1.z.enum(["child", "teen", "young-adult", "adult", "senior"]).optional(),
+    accent: zod_1.z.string().optional(),
+    language: zod_1.z.string().default("en"),
+    /** Reference to a TTS voice id once selected. */
+    voiceId: zod_1.z.string().optional(),
+    /** AG-31: emotion / delivery style. Mapped to provider voice_settings by executor. */
+    style: exports.VoiceStyleSchema.optional(),
+});
+exports.MusicProfileSchema = zod_1.z.object({
+    moods: zod_1.z
+        .array(zod_1.z.enum([
+        "energetic",
+        "calm",
+        "cinematic",
+        "uplifting",
+        "tense",
+        "dreamy",
+        "dark",
+        "playful",
+    ]))
+        .default([]),
+    bpmMin: zod_1.z.number().int().positive().optional(),
+    bpmMax: zod_1.z.number().int().positive().optional(),
+});
+exports.ClaimsSchema = zod_1.z.object({
+    /** Phrases the agent should preferentially use (brand voice). */
+    allow: zod_1.z.array(zod_1.z.string()).default([]),
+    /** Phrases the agent must never produce (legal / compliance). */
+    block: zod_1.z.array(zod_1.z.string()).default([]),
+});
+exports.ReferenceLinkSchema = zod_1.z.object({
+    url: zod_1.z.string().url(),
+    /** Set after analyze_reference runs. */
+    analyzed: zod_1.z
+        .object({
+        durationMs: zod_1.z.number(),
+        sceneCount: zod_1.z.number().int(),
+        hookMs: zod_1.z.number(),
+        cutsPerSecond: zod_1.z.number(),
+        bpm: zod_1.z.number().optional(),
+        transcriptExcerpt: zod_1.z.string().optional(),
+    })
+        .optional(),
+});
+exports.VideoBriefSchema = zod_1.z.object({
+    /** Free-text user request, kept verbatim. */
+    rawPrompt: zod_1.z.string(),
+    goal: zod_1.z.string().optional(),
+    niche: exports.VideoNicheSchema.optional(),
+    audience: zod_1.z.string().optional(),
+    tone: exports.ToneSchema.optional(),
+    hookStyle: exports.HookStyleSchema.optional(),
+    pacing: exports.PacingSchema.optional(),
+    durationSec: zod_1.z.number().positive().optional(),
+    aspect: project_schema_1.AspectSchema.optional(),
+    voice: exports.VoiceProfileSchema.optional(),
+    music: exports.MusicProfileSchema.optional(),
+    cta: zod_1.z.string().optional(),
+    claims: exports.ClaimsSchema.default({ allow: [], block: [] }),
+    references: zod_1.z.array(exports.ReferenceLinkSchema).default([]),
+    /** Tier budget envelope; the planner respects this when assigning per-scene tiers. */
+    tierBudget: project_schema_1.TierSchema.default("T1"),
+    /**
+     * How the planner should default video scenes between the two AI-video paths:
+     *   - "indirect" (default): prefer ai-image-to-video (T2I → I2V). Usually
+     *     cheaper AND higher quality — iteration on the image is ~$0.03 vs
+     *     ~$0.65 on a re-rolled text-to-video, and I2V models preserve the
+     *     composition we already vetted instead of inventing a new one.
+     *   - "direct": prefer ai-text-to-video. Pick when the brief calls for
+     *     camera motion an image can't seed (continuous orbit, parallax dolly,
+     *     dynamic crowd action) or when the brand kit pre-vets a T2V model.
+     *   - "auto": leave the choice to the planner per scene (legacy behaviour).
+     * Per-scene strategy overrides still win; this only changes the default.
+     */
+    videoStrategy: zod_1.z.enum(["direct", "indirect", "auto"]).optional(),
+    /**
+     * AG-12/13 — visual-consistency tier.
+     *   - "C0": no Visual Bible; per-scene independent prompts (cheapest, lowest consistency).
+     *   - "C1": Visual Bible built (LLM pre-pass), descriptions injected verbatim into scene prompts. ~$0.01 added per video. DEFAULT.
+     *   - "C2": C1 + pre-render ONE reference image per bible entity, thread as inputImageUrls into every scene's image-gen call. ~$0.12 added on top of C1 for a typical UGC (1 char + 1 product + 1 env). Premium tier.
+     * Frontend should default to "C1" for paid users and surface a "Premium
+     * consistency" toggle for "C2". Free-tier users get "C0".
+     */
+    consistencyTier: zod_1.z.enum(["C0", "C1", "C2"]).optional(),
+    /** Per-axis confidence (0-1) emitted by the inference pass. Used to decide what to ask. */
+    confidence: zod_1.z.record(zod_1.z.string(), zod_1.z.number().min(0).max(1)).default({}),
+});

package/lib/schemas/index.d.ts

@@ -0,0 +1,7 @@
+export * from "./project.schema";
+export * from "./brief.schema";
+export * from "./videoPlan.schema";
+export * from "./brandKit.schema";
+export * from "./agentPersona.schema";
+export * from "./agentRunJob.schema";
+//# sourceMappingURL=index.d.ts.map

package/lib/schemas/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/schemas/index.ts"],"names":[],"mappings":"AAAA,cAAc,kBAAkB,CAAC;AACjC,cAAc,gBAAgB,CAAC;AAC/B,cAAc,oBAAoB,CAAC;AACnC,cAAc,mBAAmB,CAAC;AAClC,cAAc,uBAAuB,CAAC;AACtC,cAAc,sBAAsB,CAAC"}

package/lib/schemas/index.js

@@ -0,0 +1,22 @@
+"use strict";
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __exportStar = (this && this.__exportStar) || function(m, exports) {
+    for (var p in m) if (p !== "default" && !Object.prototype.hasOwnProperty.call(exports, p)) __createBinding(exports, m, p);
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+__exportStar(require("./project.schema"), exports);
+__exportStar(require("./brief.schema"), exports);
+__exportStar(require("./videoPlan.schema"), exports);
+__exportStar(require("./brandKit.schema"), exports);
+__exportStar(require("./agentPersona.schema"), exports);
+__exportStar(require("./agentRunJob.schema"), exports);