Integration
Sarvam Provider
By This project is a provider plugin that bridges OpenClaw with Sarvam AI API, offering strong support for Indian languages (Hindi, Tamil, Telugu, Kannada, and more) with enhanced factual accuracy.
Install
npm install -gConfiguration Example
{
"agent": {
"model": "sarvam/sarvam-m",
"providers": {
"sarvam": {
"apiKey": "${SARVAM_API_KEY}"
}
}
}
}README
# OpenClaw Sarvam Provider
[](https://www.npmjs.com/package/openclaw-sarvam-provider)
[](LICENSE)
[](https://www.typescriptlang.org/)
[](https://nodejs.org/)
[](http://makeapullrequest.com)
**Official Sarvam provider for OpenClaw** - Enterprise-grade integration with wiki_grounding, strong Indian language support, and dynamic model switching.
## ✨ Features
- 🌏 **Strong Indian Language Support** - Optimized for Hindi, Tamil, Telugu, Kannada, and more
- 📚 **Wiki Grounding** - Built-in RAG-based Wikipedia retrieval for enhanced factual accuracy
- 🔄 **Dynamic Model Switching** - Seamlessly switch between Lite, M, and Pro models
- ⚡ **Streaming Support** - Real-time streaming responses for faster user experience
- 🔒 **Enterprise Security** - Environment variable support, input validation, secure communication
- 🎯 **Multiple Models** - Choose the right model for your use case
- 🛠️ **Developer-Friendly** - TypeScript with full type definitions
- 📦 **Easy Installation** - Single command setup with OpenClaw
## 📋 Requirements
- **Node.js**: ≥22.16.0
- **OpenClaw**: ≥2026.3.14
- **Sarvam API Key**: Get one at [docs.sarvam.ai](https://docs.sarvam.ai)
## 🚀 Installation
### Quick Start
```bash
npm install -g openclaw-sarvam-provider
```
### Alternative Package Managers
```bash
# Using pnpm
pnpm add -g openclaw-sarvam-provider
# Using bun
bun add -g openclaw-sarvam-provider
# Using yarn
yarn global add openclaw-sarvam-provider
```
## ⚙️ Configuration
### Step 1: Get Your API Key
1. Visit [docs.sarvam.ai](https://docs.sarvam.ai)
2. Sign up or log in to your account
3. Navigate to API Keys section
4. Generate a new API key
5. Copy your API key
### Step 2: Configure OpenClaw
#### Option A: Environment Variable (Recommended)
```bash
export SARVAM_API_KEY="your-sarvam-api-key-here"
```
Add to your shell profile for persistence:
```bash
# For Bash
echo 'export SARVAM_API_KEY="your-sarvam-api-key-here"' >> ~/.bashrc
source ~/.bashrc
# For Zsh
echo 'export SARVAM_API_KEY="your-sarvam-api-key-here"' >> ~/.zshrc
source ~/.zshrc
```
#### Option B: Interactive Setup
```bash
openclaw setup sarvam
```
#### Option C: Configuration File
Create/edit `~/.config/openclaw/config.json`:
```json
{
"agent": {
"model": "sarvam/sarvam-m",
"providers": {
"sarvam": {
"apiKey": "${SARVAM_API_KEY}"
}
}
}
}
```
## 🎯 Usage
### Basic Usage
```bash
openclaw agent --model "sarvam/sarvam-m" --message "Tell me about Indian culture"
```
### Model Switching
The provider supports **dynamic model switching**. You can switch between models at any time:
```bash
# Quick responses (Lite)
openclaw agent --model "sarvam/sarvam-lite" --message "Quick question"
# General use (M - Default)
openclaw agent --model "sarvam/sarvam-m" --message "Standard query"
# Complex analysis (Pro)
openclaw agent --model "sarvam/sarvam-pro" --wiki-grounding true --message "Complex analysis needed"
```
### Wiki Grounding (Sarvam's Unique Feature)
Enable wiki_grounding for enhanced factual accuracy:
```bash
openclaw agent --model "sarvam/sarvam-m" --wiki-grounding true --message "What is the history of the Taj Mahal?"
```
### Configuration File Usage
```json
{
"agent": {
"model": "sarvam/sarvam-m",
"temperature": 0.7,
"top_p": 0.9,
"max_tokens": 1000,
"wiki_grounding": true
}
}
```
### Multi-Agent Setup
```json
{
"agents": {
"quick": {
"model": "sarvam/sarvam-lite",
"description": "Fast responses for simple queries"
},
"general": {
"model": "sarvam/sarvam-m",
"description": "Balanced performance for general tasks"
},
"researcher": {
"model": "sarvam/sarvam-pro",
"wiki_grounding": true,
"description": "Deep research with wiki grounding"
}
}
}
```
Usage:
```bash
openclaw agent --agent quick --message "What's the time?"
openclaw agent --agent researcher --message "Analyze the economic impact of Indian independence"
```
## 📊 Available Models
| Model ID | Name | Context | Features | Best For | Pricing |
|----------|------|---------|----------|-----------|---------|
| `sarvam/sarvam-lite` | Sarvam Lite | 16K | wiki_grounding, streaming | Quick responses, simple tasks | 💰 Cheapest |
| `sarvam/sarvam-m` | Sarvam M | 32K | wiki_grounding, streaming, Indian languages | General use, balanced performance | 💳 Medium |
| `sarvam/sarvam-pro` | Sarvam Pro | 64K | wiki_grounding, streaming, advanced reasoning, Indian languages | Complex tasks, deep analysis | 💎 Premium |
### Model Selection Guide
- **Sarvam Lite** → Real-time chat, simple Q&A, high-volume queries
- **Sarvam M** → Daily use, general conversation, standard tasks
- **Sarvam Pro** → Research, analysis, complex reasoning, important content
## 🔧 Advanced Configuration
### Custom Parameters
```bash
openclaw agent \
--model "sarvam/sarvam-m" \
--temperature 0.7 \
--top-p 0.9 \
--max-tokens 500 \
--wiki-grounding true \
--message "Explain the significance of Indian classical music"
```
### API Parameters
| Parameter | Type | Default | Range | Description |
|-----------|-------|---------|--------|-------------|
| `model` | string | `sarvam-m` | `sarvam-lite`, `sarvam-m`, `sarvam-pro` | Model selection |
| `temperature` | number | `0.2` | 0.0 - 2.0 | Controls randomness |
| `top_p` | number | `1.0` | 0.0 - 1.0 | Nucleus sampling |
| `max_tokens` | number | Model-dependent | 1 - Context limit | Maximum response length |
| `wiki_grounding` | boolean | `false` | true/false | Enable Wikipedia RAG |
| `frequency_penalty` | number | `0.0` | -2.0 - 2.0 | Reduce repetition |
| `presence_penalty` | number | `0.0` | -2.0 - 2.0 | Encourage new topics |
| `stream` | boolean | `true` | true/false | Enable streaming |
### Programmatic Usage
```typescript
import { buildSarvamProvider, type SarvamRequest } from "openclaw-sarvam-provider";
const provider = buildSarvamProvider();
const config = {
apiKey: process.env.SARVAM_API_KEY,
baseURL: "https://api.sarvam.ai",
};
const request: SarvamRequest = {
messages: [
{ role: "system", content: "You are a helpful assistant." },
{ role: "user", content: "Tell me about Indian history" },
],
model: "sarvam-m",
temperature: 0.7,
wiki_grounding: true,
};
const response = await provider.chatCompletion(config, request);
console.log(response.choices[0].message.content);
```
### Streaming Example
```typescript
import { buildSarvamProvider } from "openclaw-sarvam-provider";
const provider = buildSarvamProvider();
const config = {
apiKey: process.env.SARVAM_API_KEY,
baseURL: "https://api.sarvam.ai",
};
const request = {
messages: [{ role: "user", content: "Write about Indian cuisine" }],
model: "sarvam-m",
stream: true,
wiki_grounding: true,
};
for await (const chunk of provider.chatCompletionStream(config, request)) {
for (const choice of chunk.choices) {
if (choice.delta.content) {
process.stdout.write(choice.delta.content);
}
}
}
```
## 🌟 Indian Language Support
Sarvam excels at Indian languages:
```bash
# Hindi
openclaw agent --model "sarvam/sarvam-m" --message "मुझे भारत के बारे में बताओ"
# Tamil
openclaw agent --model "sarvam/sarvam-m" --message "இந்தியாவின் வரலாறு பற்றி சொல்லுங்கள்"
# Telugu
openclaw agent --model "sarvam/sarvam-m" --message "భారతదేశం గురించి చెప్పండి"
# Kannada
openclaw agent --model "sarvam/sarvam-m" --message "ಭಾರತದ ಬಗ್ಗೆ ಹೇಳಿ"
```
## 🔒 Security
### Security Best Practices
- ✅ **Never commit API keys** to version control
- ✅ **Use environment variables** for sensitive data
- ✅ **Rotate API keys** regularly
- ✅ **Monitor API usage** for unusual activity
- ✅ **Use HTTPS only** for communications
- ✅ **Keep dependencies updated**
### Security Features
- 🔐 Environment variable support for API keys
- 🔒 HTTPS-only communications by default
- 🛡️ Input validation and sanitization
- ⏱️ Request timeouts to prevent hanging
- ✅ Type safety with TypeScript
### Security Documentation
For detailed security information, see [SECURITY_ANALYSIS.md](SECURITY_ANALYSIS.md).
## 📚 API Reference
### Types
```typescript
interface SarvamConfig {
apiKey: string;
baseURL?: string;
model?: string;
}
interface SarvamRequest {
messages: Array<{
role: "system" | "user" | "assistant";
content: string;
}>;
model: string;
temperature?: number;
top_p?: number;
max_tokens?: number;
stream?: boolean;
wiki_grounding?: boolean;
frequency_penalty?: number;
presence_penalty?: number;
}
interface SarvamResponse {
id: string;
choices: Array<{
index: number;
message: {
role: string;
content: string;
};
finish_reason: string;
}>;
created: number;
model: string;
object: string;
usage?: {
prompt_tokens: number;
completion_tokens: number;
total_tokens: number;
};
}
```
## 🛠️ Development
### Building from Source
```bash
# Clone the repository
git clone https://github.com/your-username/openclaw-sarvam-provider.git
cd openclaw-sarvam-provider
# Install dependencies
npm install
# Build the project
npm run build
# Run tests
npm test
# Watch mode for development
npm run dev
```
### Project Structure
```
openclaw-sarvam-provider/
├── src/
│ ├── index.ts # Main plugin entry point
│ └── sarvam-provider.ts # Sarvam implementation
├── dist/ # Compiled output
├── test/ # Test files
├── package.json
├── tsconfig.json
├── README.md
├── EXAMPLES.md # Usage examples
├── MODEL_SWITCHING.md # Model switching guide
├── SECURITY_ANALYSIS.md # Security documentation
└── LICENSE
```
... (truncated)
integration
Comments
Sign in to leave a comment