@fias/create-fias-plugin 1.16.0 → 1.18.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@fias/create-fias-plugin",
-  "version": "1.16.0",
+  "version": "1.18.0",
   "publishConfig": {
     "access": "public"
   },

package/templates/default/AGENTS.md

@@ -1,4 +1,4 @@
-<!-- fias-sdk-guide-version: 2.24.0 -->
+<!-- fias-sdk-guide-version: 2.26.0 -->
 
 # FIAS Plugin Development Guide
 
@@ -583,7 +583,7 @@ generate({
 });
 ```
 
-Mode values: `'text-to-image'` (baseline), `'image-to-image'` (requires `referenceImage`), `'vector'` (SVG output). Style values: `'illustration'` (stylized / digital-art) or `'photoreal'`.
+Mode values: `'text-to-image'` (baseline), `'image-to-image'` (requires `referenceImage`), `'vector'` (SVG output). Style values: `'illustration'` (stylized / digital-art) or `'photoreal'`. Provider values for `providerRecommended`: `'openai'`, `'google'`, `'stability'`, `'bfl'`, `'recraft'`, `'xai'` — xAI's Grok Imagine Image 2.0 (`ent_modeldef_grok_imagine_image_2_0`) is text-to-image only (no reference images), in `1k`, `1.5k` or `2k` resolution.
 
 **Several reference images, and output resolution.** A model whose `maxReferenceImages` (from `useImageEntities`) is more than one takes them together in `referenceImages` — Nano Banana 2 (`ent_modeldef_nano_banana_2`) takes up to 14, the GPT Image models up to 16. There is no role per image: the model reads them in order, so the prompt says which is which. `resolution` picks one of the model's `supportedResolutions`; higher costs more (`creditsPerImageByResolution`):
 
@@ -776,7 +776,13 @@ const { translation, detectedSourceLang } = await invoke<TranslateParams, Transl
 
 Specialty capabilities expose **typed convenience hooks** generated from the platform's surface registry. Each hook bakes in a `surfaceKey` and infers params/result types from the surface's schemas.
 
-Currently shipped audio hooks: `useLyriaMusic`, `useElevenLabsTts`, `useElevenLabsMusic`, `useElevenLabsSfx`, `useElevenLabsStt`, `useElevenLabsAudioIsolation`, `useElevenLabsForcedAlignment`, `useElevenLabsVoiceChanger`, `useElevenLabsVoiceDesign`. Browse the full surface list with `npx fias-dev entities`.
+Currently shipped audio hooks: `useLyriaMusic`, `useElevenLabsTts`, `useElevenLabsMusic`, `useElevenLabsSfx`, `useElevenLabsStt`, `useElevenLabsAudioIsolation`, `useElevenLabsForcedAlignment`, `useElevenLabsVoiceChanger`, `useElevenLabsVoiceDesign`, `useGeminiTts`, `useGeminiTranscribe`. Browse the full surface list with `npx fias-dev entities`.
+
+**Gemini speech.** `useGeminiTts` speaks `text` in one of 30 named voices (`voiceName`, default `'Kore'`); an optional `style` (e.g. `'cheerful and friendly'`) shapes the delivery and is not spoken. It runs Gemini 3.8 Flash TTS; pass `entityId: 'ent_intg_gemini_tts_flash_lite00'` for the lower-cost Flash-Lite. The result is a saved 24 kHz WAV (`audioUrl`, `audioRef`, `fileId`, `durationSeconds`). `useGeminiTranscribe` takes audio the same way as `useElevenLabsStt` (`fileId` or a prior `audioRef`) and returns the transcript `text`.
+
+**Sound effects and the voice changer are billed by the second.** `useElevenLabsSfx` makes a clip `durationSeconds` long (default 5 s), and `useElevenLabsVoiceChanger` bills the length of the audio it converts; both results report `secondsBilled` next to `charactersBilled`.
+
+**Lyria 3.5.** `useLyriaMusic` runs Lyria 2 (30 s instrumental clips) unless you pass `entityId: 'ent_modeldef_lyria35'`, which makes full songs — vocals or instrumental, one MP3 each, at a length the model chooses (steer it from the prompt with section tags such as `[Verse]` and `[Chorus]`). Each song takes about 20–60 s and is billed per song. `negativePrompt`, `seed` and `durationSeconds` apply to Lyria 2 only. `result.provider` is `'google-vertex'` (Lyria 2) or `'google-gemini'` (Lyria 3.5) — an exhaustive switch on it needs both cases.
 
 A legacy `useAudioGeneration()` hook still ships for back-compat with code written before the typed-surface migration. New plugins should use the typed hooks above — the legacy hook will be removed in a future major version.
 

package/templates/default/CLAUDE.md

@@ -1,4 +1,4 @@
-<!-- fias-sdk-guide-version: 2.24.0 -->
+<!-- fias-sdk-guide-version: 2.26.0 -->
 
 # FIAS Plugin Development Guide
 
@@ -583,7 +583,7 @@ generate({
 });
 ```
 
-Mode values: `'text-to-image'` (baseline), `'image-to-image'` (requires `referenceImage`), `'vector'` (SVG output). Style values: `'illustration'` (stylized / digital-art) or `'photoreal'`.
+Mode values: `'text-to-image'` (baseline), `'image-to-image'` (requires `referenceImage`), `'vector'` (SVG output). Style values: `'illustration'` (stylized / digital-art) or `'photoreal'`. Provider values for `providerRecommended`: `'openai'`, `'google'`, `'stability'`, `'bfl'`, `'recraft'`, `'xai'` — xAI's Grok Imagine Image 2.0 (`ent_modeldef_grok_imagine_image_2_0`) is text-to-image only (no reference images), in `1k`, `1.5k` or `2k` resolution.
 
 **Several reference images, and output resolution.** A model whose `maxReferenceImages` (from `useImageEntities`) is more than one takes them together in `referenceImages` — Nano Banana 2 (`ent_modeldef_nano_banana_2`) takes up to 14, the GPT Image models up to 16. There is no role per image: the model reads them in order, so the prompt says which is which. `resolution` picks one of the model's `supportedResolutions`; higher costs more (`creditsPerImageByResolution`):
 
@@ -776,7 +776,13 @@ const { translation, detectedSourceLang } = await invoke<TranslateParams, Transl
 
 Specialty capabilities expose **typed convenience hooks** generated from the platform's surface registry. Each hook bakes in a `surfaceKey` and infers params/result types from the surface's schemas.
 
-Currently shipped audio hooks: `useLyriaMusic`, `useElevenLabsTts`, `useElevenLabsMusic`, `useElevenLabsSfx`, `useElevenLabsStt`, `useElevenLabsAudioIsolation`, `useElevenLabsForcedAlignment`, `useElevenLabsVoiceChanger`, `useElevenLabsVoiceDesign`. Browse the full surface list with `npx fias-dev entities`.
+Currently shipped audio hooks: `useLyriaMusic`, `useElevenLabsTts`, `useElevenLabsMusic`, `useElevenLabsSfx`, `useElevenLabsStt`, `useElevenLabsAudioIsolation`, `useElevenLabsForcedAlignment`, `useElevenLabsVoiceChanger`, `useElevenLabsVoiceDesign`, `useGeminiTts`, `useGeminiTranscribe`. Browse the full surface list with `npx fias-dev entities`.
+
+**Gemini speech.** `useGeminiTts` speaks `text` in one of 30 named voices (`voiceName`, default `'Kore'`); an optional `style` (e.g. `'cheerful and friendly'`) shapes the delivery and is not spoken. It runs Gemini 3.8 Flash TTS; pass `entityId: 'ent_intg_gemini_tts_flash_lite00'` for the lower-cost Flash-Lite. The result is a saved 24 kHz WAV (`audioUrl`, `audioRef`, `fileId`, `durationSeconds`). `useGeminiTranscribe` takes audio the same way as `useElevenLabsStt` (`fileId` or a prior `audioRef`) and returns the transcript `text`.
+
+**Sound effects and the voice changer are billed by the second.** `useElevenLabsSfx` makes a clip `durationSeconds` long (default 5 s), and `useElevenLabsVoiceChanger` bills the length of the audio it converts; both results report `secondsBilled` next to `charactersBilled`.
+
+**Lyria 3.5.** `useLyriaMusic` runs Lyria 2 (30 s instrumental clips) unless you pass `entityId: 'ent_modeldef_lyria35'`, which makes full songs — vocals or instrumental, one MP3 each, at a length the model chooses (steer it from the prompt with section tags such as `[Verse]` and `[Chorus]`). Each song takes about 20–60 s and is billed per song. `negativePrompt`, `seed` and `durationSeconds` apply to Lyria 2 only. `result.provider` is `'google-vertex'` (Lyria 2) or `'google-gemini'` (Lyria 3.5) — an exhaustive switch on it needs both cases.
 
 A legacy `useAudioGeneration()` hook still ships for back-compat with code written before the typed-surface migration. New plugins should use the typed hooks above — the legacy hook will be removed in a future major version.