@elizaos/plugin-google-genai 2.0.0-beta.1 → 2.0.3-beta.2

package/README.md

@@ -1,71 +1,82 @@
 # @elizaos/plugin-google-genai
 
-Google Generative AI (Gemini) plugin for elizaOS with native support for TypeScript, Python, and Rust.
+Google Generative AI (Gemini) model provider for [elizaOS](https://github.com/elizaos/eliza) agents. Registers handlers for text generation, embeddings, and image description across all elizaOS model tiers, backed by the Google Generative AI API.
 
-## Features
+## Capabilities
 
-- **Text Generation**: Generate text using Gemini 2.0 Flash and 2.5 Pro models
-- **Embeddings**: Generate text embeddings with text-embedding-004
-- **Image Analysis**: Analyze and describe images with multimodal capabilities
-- **JSON Object Generation**: Generate structured JSON with schema validation
+- **Text generation** across all model tiers: nano, small, medium, large, mega, response handler, action planner.
+- **Text embeddings** with `text-embedding-004` (768 dimensions).
+- **Image description** — fetch an image by URL, encode it inline, and return a `{ title, description }` object.
+- **Structured output** — pass a JSON Schema as `responseSchema` to any text handler to get `application/json` back from the model.
+- **Tool use** — pass function declarations via `tools` / `toolChoice` to enable function-calling on supported models.
 
-## Available Models
+## Auto-enable
 
-| Model Type        | Default Model                | Description                        |
-| ----------------- | ---------------------------- | ---------------------------------- |
-| TEXT_SMALL        | gemini-2.0-flash-001         | Fast text + structured output (responseSchema, tools) |
-| TEXT_LARGE        | gemini-2.5-pro-preview-03-25 | Capable text + structured output (responseSchema, tools) |
-| TEXT_EMBEDDING    | text-embedding-004           | Text embeddings (768 dimensions)   |
-| IMAGE_DESCRIPTION | gemini-2.5-pro-preview-03-25 | Multimodal image analysis          |
+The plugin is automatically enabled by elizaOS when any of the following environment variables is set and non-empty:
 
-## Installation
+- `GOOGLE_API_KEY`
+- `GOOGLE_GENERATIVE_AI_API_KEY`
+- `GEMINI_API_KEY`
 
-### TypeScript/JavaScript (npm)
+## Installation
 
 ```bash
-npm install @elizaos/plugin-google-genai
-# or
 bun add @elizaos/plugin-google-genai
 ```
-## Configuration
 
-Set the following environment variables:
+Or register it explicitly in your agent character file:
+
+```json
+{
+  "plugins": ["@elizaos/plugin-google-genai"]
+}
+```
+
+## Configuration
 
-| Variable                       | Required | Description                                                                  |
-| ------------------------------ | -------- | ---------------------------------------------------------------------------- |
-| `GOOGLE_GENERATIVE_AI_API_KEY` | Yes      | Your Google AI API key from [Google AI Studio](https://aistudio.google.com/) |
-| `GOOGLE_SMALL_MODEL`           | No       | Override small model (default: gemini-2.0-flash-001)                         |
-| `GOOGLE_LARGE_MODEL`           | No       | Override large model (default: gemini-2.5-pro-preview-03-25)                 |
-| `GOOGLE_EMBEDDING_MODEL`       | No       | Override embedding model (default: text-embedding-004)                       |
-| `GOOGLE_IMAGE_MODEL`           | No       | Override image analysis model                                                |
-| `GOOGLE_TIMEOUT_SECONDS`       | No       | Request timeout (default: 60)                                                |
+| Environment variable | Required | Default | Description |
+|---|---|---|---|
+| `GOOGLE_GENERATIVE_AI_API_KEY` | Yes | — | API key from [Google AI Studio](https://aistudio.google.com/) |
+| `GOOGLE_SMALL_MODEL` | No | `gemini-2.0-flash-001` | Small/fast text model |
+| `GOOGLE_LARGE_MODEL` | No | `gemini-2.5-pro-preview-03-25` | Large/capable text model |
+| `GOOGLE_NANO_MODEL` | No | falls back to small | Nano text model |
+| `GOOGLE_MEDIUM_MODEL` | No | falls back to small | Medium text model |
+| `GOOGLE_MEGA_MODEL` | No | falls back to large | Mega text model |
+| `GOOGLE_RESPONSE_HANDLER_MODEL` | No | falls back to nano | Response handler model |
+| `GOOGLE_ACTION_PLANNER_MODEL` | No | falls back to medium | Action planner model |
+| `GOOGLE_EMBEDDING_MODEL` | No | `text-embedding-004` | Embedding model |
+| `GOOGLE_IMAGE_MODEL` | No | `gemini-2.5-pro-preview-03-25` | Image description model |
+
+Generic fallbacks (`SMALL_MODEL`, `LARGE_MODEL`, `IMAGE_MODEL`, etc.) are also respected when the `GOOGLE_*` prefix variants are not set.
 
 ## Usage
 
-### TypeScript (elizaOS Plugin)
+Once the plugin is loaded, use any Gemini model through the standard elizaOS runtime interface:
 
 ```typescript
-import { googleGenAIPlugin } from "@elizaos/plugin-google-genai";
+import { ModelType } from "@elizaos/core";
 
-// Register the plugin with your elizaOS agent
-const agent = new Agent({
-  plugins: [googleGenAIPlugin],
-});
-
-// Use via runtime
+// Text generation
 const text = await runtime.useModel(ModelType.TEXT_LARGE, {
-  prompt: "Explain quantum mechanics in simple terms.",
+  prompt: "Explain quantum entanglement in plain language.",
 });
 
+// Embeddings
 const embedding = await runtime.useModel(ModelType.TEXT_EMBEDDING, {
   text: "Hello, world!",
 });
-
-// Structured output: route through TEXT_* with `responseSchema`. Google's
-// SDK accepts the schema via `responseJsonSchema` + `responseMimeType` set
-// internally by the handler.
-const structured = await runtime.useModel(ModelType.TEXT_SMALL, {
-  prompt: "Generate a person profile with name and age.",
+// embedding is number[] with 768 dimensions
+
+// Image description
+const result = await runtime.useModel(
+  ModelType.IMAGE_DESCRIPTION,
+  "https://example.com/image.jpg",
+);
+// result: { title: string; description: string }
+
+// Structured output
+const person = await runtime.useModel(ModelType.TEXT_SMALL, {
+  prompt: "Generate a sample person profile.",
   responseSchema: {
     type: "object",
     properties: {
@@ -76,49 +87,28 @@ const structured = await runtime.useModel(ModelType.TEXT_SMALL, {
   },
 });
 ```
-## Project Structure
 
-```
-plugin-google-genai/
-├── typescript/          # TypeScript implementation
-│   ├── index.ts         # Main plugin entry
-│   ├── models/          # Model handlers
-│   ├── utils/           # Utility functions
-│   └── __tests__/       # Unit and integration tests
-├── package.json         # npm publishing config
-└── README.md            # This file
-```
+## Available model tiers
 
-## Development
+| ModelType | Default model | Notes |
+|---|---|---|
+| `TEXT_NANO` | falls back to small | Fastest; shares small model by default |
+| `TEXT_SMALL` | `gemini-2.0-flash-001` | Fast + structured output |
+| `TEXT_MEDIUM` | falls back to small | |
+| `TEXT_LARGE` | `gemini-2.5-pro-preview-03-25` | High-quality + structured output |
+| `TEXT_MEGA` | falls back to large | |
+| `RESPONSE_HANDLER` | falls back to nano | |
+| `ACTION_PLANNER` | falls back to medium | |
+| `TEXT_EMBEDDING` | `text-embedding-004` | 768-dim vectors |
+| `IMAGE_DESCRIPTION` | `gemini-2.5-pro-preview-03-25` | Multimodal; fetches image by URL |
 
-### TypeScript
+## Development
 
 ```bash
-# Install dependencies
 bun install
-
-# Build
-bun run build
-
-# Run tests
-bun run test
-
-# Type checking
-bun run typecheck
-```
-
-## Publishing
-
-### npm (TypeScript)
-
-```bash
-npm publish --access public
+bun run --cwd plugins/plugin-google-genai build
+bun run --cwd plugins/plugin-google-genai test
+bun run --cwd plugins/plugin-google-genai typecheck
 ```
 
-## License
-
-MIT
-
-## Contributing
-
-See the main [elizaOS contribution guidelines](https://github.com/elizaos/eliza/blob/main/CONTRIBUTING.md).
+See [AGENTS.md](AGENTS.md) for the agent-facing layout reference and extension guide.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@elizaos/plugin-google-genai",
-  "version": "2.0.0-beta.1",
+  "version": "2.0.3-beta.2",
   "type": "module",
   "main": "dist/node/index.node.js",
   "module": "dist/node/index.node.js",
@@ -28,6 +28,12 @@
         "default": "./dist/node/index.node.js"
       },
       "default": "./dist/node/index.node.js"
+    },
+    "./*.css": "./dist/*.css",
+    "./*": {
+      "types": "./dist/*.d.ts",
+      "import": "./dist/*.js",
+      "default": "./dist/*.js"
     }
   },
   "files": [
@@ -52,7 +58,7 @@
   },
   "devDependencies": {
     "@biomejs/biome": "^2.4.14",
-    "@elizaos/core": "2.0.0-beta.1",
+    "@elizaos/core": "2.0.3-beta.2",
     "@types/node": "^25.0.3",
     "bun-types": "^1.3.12",
     "dotenv": "^17.2.3",
@@ -60,14 +66,14 @@
     "vitest": "^4.0.0"
   },
   "peerDependencies": {
-    "@elizaos/core": "2.0.0-beta.1"
+    "@elizaos/core": "2.0.3-beta.2"
   },
   "scripts": {
     "dev": "bun run build.ts --watch",
     "clean": "rm -rf dist .turbo .turbo-tsconfig.json tsconfig.tsbuildinfo",
     "format": "bunx @biomejs/biome format --write .",
     "format:check": "bunx @biomejs/biome format .",
-    "typecheck": "tsc --noEmit -p tsconfig.json",
+    "typecheck": "tsgo --noEmit -p tsconfig.json",
     "test": "bunx vitest run --config vitest.config.ts",
     "test:unit": "bunx vitest run --config vitest.config.ts --dir __tests__/unit",
     "test:integration": "bunx vitest run --config vitest.config.ts --dir __tests__/integration",
@@ -140,7 +146,7 @@
       }
     }
   },
-  "gitHead": "646c632924826e2b75c2304a75ee56959fe4a460",
+  "gitHead": "82fe0f44215954c2417328203f5bd6510985c1fc",
   "eliza": {
     "platforms": [
       "browser",