Relayplane

https://relayplane.com/docs
Admin
Relayplane is an intelligent API proxy that automatically routes LLM requests to optimal models,...
Tokens:112,518
Snippets:959
Trust Score:5.2
Update:6 months ago
Show doc for...
Tokens:
Raw
### Quick Start RelayPlane Proxy Setup

Source: https://relayplane.com/docs/proxy/cli

A step-by-step guide to start the RelayPlane proxy, configure AI agent environment variables to point to it, and verify savings.

```bash
export ANTHROPIC_API_KEY=sk-ant-api-...
relayplane
```

```bash
export ANTHROPIC_BASE_URL=http://localhost:4100
export OPENAI_BASE_URL=http://localhost:4100
```

```bash
claude  # or cursor, aider, etc.
```

```bash
relayplane stats
```

--------------------------------

### Install and start RelayPlane via npm

Source: https://relayplane.com/docs/installation

Installs the CLI globally and initializes the proxy service.

```bash
npm install -g @relayplane/proxy
relayplane start
```

--------------------------------

### Quick Start

Source: https://relayplane.com/docs/sdk

A quick start guide demonstrating how to configure providers and run a basic workflow using the Relayplane SDK.

```APIDOC
## Quick Start

```typescript
1import { relay } from '@relayplane/sdk';
2
3// Configure providers
4relay.configure({
5  providers: {
6    openai: { apiKey: process.env.OPENAI_API_KEY! },
7    anthropic: { apiKey: process.env.ANTHROPIC_API_KEY! }
8  }
9});
10
11// Create and run a workflow
12const result = await relay
13  .workflow('my-workflow')
14  .step('extract', {
15    schema: MySchema,
16    systemPrompt: 'Extract data from the input'
17  })
18  .with('openai:gpt-4o')
19  .step('summarize')
20  .with('anthropic:claude-3.5-sonnet')
21  .depends('extract')
22  .run({ input: 'Your data here' });
23
24console.log(result.steps.summarize);
```
```

--------------------------------

### Install and Start Relayplane Proxy

Source: https://relayplane.com/docs/quickstart

Install the Relayplane proxy globally and then start the proxy service. Node.js 18+ is required.

```bash
1npm install -g @relayplane/proxy
2relayplane start
```

--------------------------------

### Install and Start RelayPlane Proxy

Source: https://relayplane.com/docs

Install the RelayPlane proxy globally using npm and then start the proxy service. Point your AI tools to the running proxy by setting the appropriate environment variables for the API base URL.

```bash
# Install the proxy
npm install -g @relayplane/proxy

# Start it
relayplane

# Point your tools at it
export ANTHROPIC_BASE_URL=http://localhost:4100
export OPENAI_BASE_URL=http://localhost:4100
```

--------------------------------

### Quick Start Lead Qualification

Source: https://relayplane.com/docs/examples/lead-qualifier

A minimal example demonstrating how to initialize a lead-qualifier workflow with enrichment and scoring steps.

```typescript
import { relay } from "@relayplane/sdk";

const result = await relay
  .workflow("lead-qualifier")
  .step("enrich").with("openai:gpt-4o")
  .step("score").with("anthropic:claude-3.5-sonnet").depends("enrich")
  .run({ email: "prospect@company.com", company: "Acme Corp" });
```

--------------------------------

### Install Node.js on Debian/Ubuntu

Source: https://relayplane.com/docs/vps-deployment

Uses the NodeSource setup script to install Node 20 LTS.

```bash
1# Install curl if needed
2sudo apt-get update && sudo apt-get install -y curl
3
4# Add NodeSource repo for Node 20
5curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash -
6
7# Install Node.js
8sudo apt-get install -y nodejs
9
10# Verify
11node --version   # v20.x.x
12npm --version
```

--------------------------------

### Installation

Source: https://relayplane.com/docs/cli

Instructions for installing the RelayPlane CLI, either with the SDK or as a standalone package.

```APIDOC
## Installation
The CLI is included with the SDK, or install it separately:

```bash
# Install with SDK (recommended)
npm install @relayplane/sdk

# Or install CLI only
npm install -g @relayplane/proxy
```
```

--------------------------------

### Install @relayplane/sdk

Source: https://relayplane.com/docs/sdk

Install the SDK using npm, pnpm, or yarn.

```bash
1npm install @relayplane/sdk
2# or
3pnpm add @relayplane/sdk
4# or
5yarn add @relayplane/sdk
```

--------------------------------

### Installation

Source: https://relayplane.com/docs/sdk

Instructions for installing the @relayplane/sdk package using npm, pnpm, or yarn.

```APIDOC
## Installation

```bash
1npm install @relayplane/sdk
2# or
3pnpm add @relayplane/sdk
4# or
5yarn add @relayplane/sdk
```
```

--------------------------------

### Verify RelayPlane installation

Source: https://relayplane.com/docs/installation

Checks the installed version of the CLI to confirm successful setup.

```bash
relayplane --version
```

--------------------------------

### Quick Contract Analysis Example

Source: https://relayplane.com/docs/examples/contract-analyzer

A concise example demonstrating how to initiate the contract analysis workflow with two steps: extracting terms and identifying risks.

```typescript
import { relay } from "@relayplane/sdk";

const result = await relay
  .workflow("contract-analyzer")
  .step("extract-terms").with("anthropic:claude-3.5-sonnet")
  .step("identify-risks").with("openai:gpt-4o").depends("extract-terms")
  .run({ contractText: "..." });
```

--------------------------------

### Install nginx

Source: https://relayplane.com/docs/vps-deployment

Install the nginx web server on a Debian-based system.

```bash
sudo apt-get install -y nginx
```

--------------------------------

### Install RelayPlane CLI

Source: https://relayplane.com/docs/cli

Install the RelayPlane CLI using npm. It can be installed with the SDK or separately as a global package.

```bash
npm install @relayplane/sdk
```

```bash
npm install -g @relayplane/proxy
```

--------------------------------

### Install Relayplane Proxy for API Key Users

Source: https://relayplane.com/docs/quickstart

Install the Relayplane proxy globally. This is the first step for users who manage their API keys directly.

```bash
1npm install -g @relayplane/proxy
```

--------------------------------

### Install RelayPlane Proxy

Source: https://relayplane.com/docs/vps-deployment

Installs the proxy globally via npm.

```bash
1sudo npm install -g @relayplane/proxy
2
3# Verify the binary is available
4relayplane --version
```

--------------------------------

### Complete Workflow with Multiple Parallel Branches

Source: https://relayplane.com/docs/features/parallel

This comprehensive example showcases a workflow with two independent starting points ('extract-text', 'extract-images') that run in parallel. It then demonstrates parallel analysis of text and images, followed by a final combination step.

```javascript
import { relay } from "@relayplane/sdk";

const result = await relay
  .workflow("multi-analysis")

  // Two independent starting points (run in parallel)
  .step("extract-text")
  .with("openai:gpt-4o")
  .prompt("Extract text from document")

  .step("extract-images")
  .with("openai:gpt-4o-vision")
  .prompt("Extract and describe images")

  // Parallel analysis of text
  .step("text-sentiment")
  .with("anthropic:claude-sonnet-4-20250514")
  .depends("extract-text")

  .step("text-topics")
  .with("openai:gpt-4o")
  .depends("extract-text")

  // Image analysis (parallel with text analysis)
  .step("image-classification")
  .with("openai:gpt-4o-vision")
  .depends("extract-images")

  // Final combination
  .step("final-report")
  .with("openai:gpt-4o")
  .depends("text-sentiment", "text-topics", "image-classification")
  .prompt("Generate comprehensive report")

  .run({ document: file });

// Check execution timing
console.log(`Total duration: ${result.durationMs}ms`);
for (const step of result.steps) {
  console.log(`  ${step.stepName}: ${step.durationMs}ms`);
}
```

--------------------------------

### Start Relayplane Proxy for API Key Users

Source: https://relayplane.com/docs/quickstart

Start the Relayplane proxy service. This command is used after setting up API keys and the base URL.

```bash
1relayplane start
```

--------------------------------

### Environment File Content

Source: https://relayplane.com/docs/vps-deployment

Example configuration file for API keys and proxy settings.

```text
1# AI Provider Keys (add the ones you use)
2ANTHROPIC_API_KEY=sk-ant-...
3OPENAI_API_KEY=sk-...
4GOOGLE_AI_API_KEY=...
5XAI_API_KEY=...
6
7# Proxy port (nginx will forward to this)
8RELAYPLANE_PORT=4000
9
10# Optional: your RelayPlane cloud API key for telemetry
11RELAYPLANE_API_KEY=rp-...
12
13NODE_ENV=production
14LOG_LEVEL=info
```

--------------------------------

### Structured Logging Setup with Pino

Source: https://relayplane.com/docs/deployment

Configure structured logging using the Pino library for production monitoring. This setup includes custom log levels, ISO timestamps, and request context mixins for enhanced observability.

```typescript
import pino from "pino";

export const logger = pino({
  level: process.env.LOG_LEVEL || "info",
  formatters: {
    level: (label) => ({ level: label }),
  },
  timestamp: pino.stdTimeFunctions.isoTime,
  // Add request context
  mixin() {
    return {
      service: "relayplane-workflows",
      environment: process.env.NODE_ENV,
    };
  },
});

// Usage in workflows
logger.info({ workflowName: "invoice-processor", duration: 1234 }, "Workflow completed");
logger.error({ error: err.message, stack: err.stack }, "Workflow failed");
```

--------------------------------

### Install Ollama and Pull Models

Source: https://relayplane.com/docs/providers

Instructions for installing Ollama on macOS/Linux and pulling specific AI models. This is a prerequisite for local model execution.

```bash
# Install Ollama (macOS/Linux)
curl -fsSL https://ollama.com/install.sh | sh

# Pull models
ollama pull llama3.3
ollama pull qwen2.5
ollama pull deepseek-r1
ollama pull mistral

# Start Ollama server (runs on http://localhost:11434)
ullama serve
```

--------------------------------

### Install ClawHub Skill

Source: https://relayplane.com/docs/integrations/openclaw

Install the RelayPlane skill for OpenClaw using the ClawHub package manager.

```bash
1clawhub install relayplane
```

--------------------------------

### Start Relayplane Proxy

Source: https://relayplane.com/docs/reference/errors

Command to start the proxy service when encountering connection refused errors.

```bash
relayplane start
```

--------------------------------

### Example RelayPlane Config JSON

Source: https://relayplane.com/docs/proxy/configuration

This is an example of a complete RelayPlane configuration file in JSON format. It includes settings for device ID, telemetry, API key, and mesh learning.

```json
{
  "device_id": "anon_a1b2c3d4e5f6g7h8",
  "telemetry_enabled": false,
  "first_run_complete": true,
  "config_version": 1,
  "created_at": "2026-02-25T18:00:00.000Z",
  "updated_at": "2026-02-25T18:10:00.000Z",
  "mesh": {
    "enabled": false,
    "endpoint": "https://osmosis-mesh-dev.fly.dev",
    "sync_interval_ms": 60000,
    "contribute": true
  }
}
```

--------------------------------

### RelayPlane Installation and Configuration

Source: https://relayplane.com/docs/mcp-server/installation

Instructions for setting up the RelayPlane MCP server in various environments and configuring provider API keys.

```APIDOC
## Installation

### Claude Code
Add the server using the following command:
`claude mcp add relayplane -- npx @relayplane/mcp-server`

### Cursor
Add to your Cursor MCP configuration file:
```json
{
  "mcpServers": {
    "relayplane": {
      "command": "npx",
      "args": ["@relayplane/mcp-server"],
      "env": {
        "OPENAI_API_KEY": "sk-...",
        "ANTHROPIC_API_KEY": "sk-ant-..."
      }
    }
  }
}
```

### Environment Variables
- **OPENAI_API_KEY** (string) - Required - OpenAI API key
- **ANTHROPIC_API_KEY** (string) - Required - Anthropic API key
- **GOOGLE_API_KEY** (string) - Optional - Google API key
- **XAI_API_KEY** (string) - Optional - xAI API key
```

--------------------------------

### Basic CLI Usage

Source: https://relayplane.com/docs/proxy/cli

Use `npx` to run the relayplane CLI or use the installed command directly.

```bash
npx @relayplane/proxy [command] [options]
```

```bash
relayplane [command] [options]
```

--------------------------------

### Sample Release Changelog Output

Source: https://relayplane.com/docs/examples/pr-changelog

An example of a changelog generated for release v3.2.0, detailing breaking changes, new features, bug fixes, performance improvements, documentation updates, and contributors. Includes a diff example for a breaking change.

```markdown
# Release v3.2.0 - 2024-11-18

## ⚠️ Breaking Changes

**API Authentication now requires Bearer tokens**
Previously, API keys could be passed via query parameters. For security, we now require authentication via the `Authorization` header.

Migration:
```diff
- fetch('https://api.example.com/data?api_key=xxx')
+ fetch('https://api.example.com/data', {
+   headers: { 'Authorization': 'Bearer xxx' }
+ })
```

Timeline: Query param auth deprecated immediately, removed in v4.0.0 (Jan 2025)

## 🚀 New Features

- **Multi-model workflows**: Chain steps across OpenAI, Anthropic, and Google models in a single workflow (#247)
- **Webhook retry logic**: Automatically retry failed webhook deliveries with exponential backoff (#251)
- **Team workspaces**: Collaborate on workflows with team members and role-based permissions (#256)

## 🐛 Bug Fixes

- Fixed workflow execution timeout not being respected (#243)
- Resolved memory leak in long-running scheduled jobs (#248)
- Corrected timezone handling for cron schedules (#253)

## ⚡ Performance Improvements

- Reduced workflow cold start time by 60% through better caching (#245)
- Database query optimization decreased API latency by 200ms (#249)

## 📚 Documentation

- Added 25 production-ready workflow examples (#258)
- Created migration guide for v2 to v3 upgrade (#259)

## 🙏 Contributors

Thank you to @sarah-dev, @mike-engineer, @alex-designer, and 12 other contributors for making this release possible!

**Full Changelog**: https://github.com/org/repo/compare/v3.1.0...v3.2.0
```

--------------------------------

### Start Claude Code

Source: https://relayplane.com/docs/quickstart

Launch Claude Code. Requests will automatically flow through the configured proxy.

```bash
1claude
```

--------------------------------

### Quick Start: Configure and Run a Workflow

Source: https://relayplane.com/docs/sdk

Configure providers with API keys and then create and run a workflow with multiple steps, specifying models for each step.

```typescript
1import { relay } from '@relayplane/sdk';
2
3// Configure providers
4relay.configure({
  providers: {
    openai: { apiKey: process.env.OPENAI_API_KEY! },
    anthropic: { apiKey: process.env.ANTHROPIC_API_KEY! }
  }
});
10
11// Create and run a workflow
12const result = await relay
13  .workflow('my-workflow')
14  .step('extract', {
15    schema: MySchema,
16    systemPrompt: 'Extract data from the input'
17  })
18  .with('openai:gpt-4o')
19  .step('summarize')
20  .with('anthropic:claude-3.5-sonnet')
21  .depends('extract')
22  .run({ input: 'Your data here' });
23
24console.log(result.steps.summarize);
```

--------------------------------

### Manage Autostart

Source: https://relayplane.com/docs/proxy/cli

Legacy command for managing autostart on Linux systems using systemd.

```bash
relayplane autostart status
sudo relayplane autostart on
sudo relayplane autostart off
```

--------------------------------

### Configure Anthropic Provider

Source: https://relayplane.com/docs/providers

Integrate the Anthropic provider by setting your API key. This setup is typically performed once when your application starts.

```typescript
import { relay } from '@relayplane/sdk'

// Configure your provider (do this once at app startup)
relay.configure({
  providers: {
    anthropic: { apiKey: process.env.ANTHROPIC_API_KEY! }
  }
})

const analysisPrompt = `You are an expert analyst. Thoroughly analyze
the following document and provide detailed insights with
supporting evidence from the text.

Document: {{input.document}}`

const workflow = relay
  .workflow('anthropic-example')
  .step('deep-analysis')
  .with('anthropic:claude-sonnet-4-5-20250929')
  .prompt(analysisPrompt)

// longDocument can be up to 200k tokens
const result = await workflow.run({ document: longDocument })
```

--------------------------------

### Complete Relayplane SDK Example

Source: https://relayplane.com/docs/features/fallback

Configure multiple providers, define a schema, build a resilient workflow with fallbacks, and run a prompt. Ensure API keys are set in environment variables.

```typescript
import { relay } from "@relayplane/sdk";
import { z } from "zod";

// Configure multiple providers
relay.configure({
  providers: {
    openai: { apiKey: process.env.OPENAI_API_KEY },
    anthropic: { apiKey: process.env.ANTHROPIC_API_KEY },
  }
});

// Define schema for structured extraction
const SentimentSchema = z.object({
  sentiment: z.enum(["positive", "negative", "neutral"]),
  confidence: z.number().min(0).max(1),
  keywords: z.array(z.string()),
});

// Build resilient workflow with fallbacks
const result = await relay
  .workflow("sentiment-analysis")
  .step("analyze", {
    schema: SentimentSchema,
    retry: { maxRetries: 2, backoffMs: 500 }
  })
  .with("openai:gpt-4o")
  .fallback("anthropic:claude-sonnet-4-20250514")
  .fallback("openai:gpt-4o-mini")
  .prompt(`Analyze the sentiment of this text:

Text: {{input.text}}

Return sentiment (positive/negative/neutral), confidence score, and key phrases.`)
  .run({ text: "I absolutely love this product! Best purchase ever." });

if (result.success) {
  console.log("Sentiment:", result.finalOutput.sentiment);
  console.log("Confidence:", result.finalOutput.confidence);
} else {
  console.error("All models failed:", result.error?.message);
}
```

--------------------------------

### Enable nginx Configuration

Source: https://relayplane.com/docs/vps-deployment

Activate the site configuration and verify the syntax before reloading.

```bash
# Enable the site and reload nginx
sudo ln -s /etc/nginx/sites-available/relayplane /etc/nginx/sites-enabled/relayplane
sudo nginx -t   # test config
sudo systemctl reload nginx
```

--------------------------------

### Manage Configuration

Source: https://relayplane.com/docs/proxy/cli

View the current configuration or set the API key for Pro features.

```bash
relayplane config             # Show current config
relayplane config set-key rp_sk_abc123  # Set API key
```

--------------------------------

### Typical RelayPlane Workflow Setup

Source: https://relayplane.com/docs/cli

A sequence of commands to set up and manage RelayPlane workflows, including login, status check, listing workflows, syncing to the cloud, and opening the dashboard.

```bash
npx relay login
npx relay status
npx relay workflows
npx relay sync
npx relay dashboard
```

--------------------------------

### Start RelayPlane on a Different Port

Source: https://relayplane.com/docs/faq

If port 4100 is already in use, you can start the RelayPlane proxy on an alternative port.

```bash
relayplane start --port 3002
```

--------------------------------

### Configure Environment Variables

Source: https://relayplane.com/docs/vps-deployment

Creates the configuration directory and environment file.

```bash
1sudo mkdir -p /etc/relayplane
2sudo nano /etc/relayplane/.env
```

--------------------------------

### Main Exports: Relay Workflow and Configuration

Source: https://relayplane.com/docs/sdk

Demonstrates creating a workflow, configuring the SDK globally, retrieving the current configuration, and resetting the configuration for testing.

```typescript
1import { relay } from '@relayplane/sdk';
2
3// Create a workflow
4const workflow = relay.workflow('name');
5
6// Configure globally
7relay.configure({ ... });
8
9// Get current config
10const config = relay.getConfig();
11
12// Reset config (for testing)
13relay.resetConfig();
```

--------------------------------

### Build RelayPlane from source

Source: https://relayplane.com/docs/installation

Clones the repository and builds the project locally.

```bash
git clone https://github.com/RelayPlane/proxy.git
cd proxy
npm install
npm run build
npm link
```

--------------------------------

### Start RelayPlane Proxy

Source: https://relayplane.com/docs/proxy/cli

Start the proxy server on the default port 4100, listening only on localhost. Custom ports and host bindings are supported. Audit and offline modes can be enabled.

```bash
# Start on default port 4100 (localhost-only, safest)
relayplane
```

```bash
# Explicit start command (equivalent)
relayplane start
```

```bash
# Localhost-first custom binding
relayplane --port 3002 --host 127.0.0.1
```

```bash
# Network-exposed binding (only behind firewall/reverse proxy/auth)
relayplane --port 3002 --host 0.0.0.0
```

```bash
# Audit mode — see telemetry before it's sent
relayplane --audit
```

```bash
# Offline mode — no telemetry, no cloud sync
relayplane --offline
```

```bash
# Verbose logging
relayplane -v
```

--------------------------------

### Complete Workflow Example with Multiple Retries

Source: https://relayplane.com/docs/features/retry

Demonstrates a complete RelayPlane workflow with different retry configurations for multiple steps. It also shows how to check if any steps were retried during execution.

```javascript
import { relay } from "@relayplane/sdk";

const result = await relay
  .workflow("resilient-extraction")
  .step("extract", {
    retry: { maxRetries: 3, backoffMs: 1000 },
    systemPrompt: "Extract key information as JSON"
  })
  .with("openai:gpt-4o")
  .step("validate", {
    retry: { maxRetries: 2, backoffMs: 500 },
    systemPrompt: "Validate the extracted data"
  })
  .with("anthropic:claude-sonnet-4-20250514")
  .depends("extract")
  .run({ document: "..." });

// Check if any retries occurred
for (const step of result.steps) {
  if (step.retryCount > 0) {
    console.log(`Step ${step.stepName} retried ${step.retryCount} times`);
  }
}
```

--------------------------------

### GET /v1/telemetry/runs

Source: https://relayplane.com/docs/cost-optimization/usage-tracking

Retrieves the most recent request history.

```APIDOC
## GET /v1/telemetry/runs

### Description
Retrieves recent request history.

### Method
GET

### Endpoint
/v1/telemetry/runs

### Parameters
#### Query Parameters
- **limit** (integer) - Optional - Maximum number of entries to return.
```

--------------------------------

### Configure and Run a Basic MCP Workflow

Source: https://relayplane.com/docs/mcp/quickstart

Demonstrates the full lifecycle of a workflow, including provider configuration, step definition with AI and MCP tools, and execution.

```typescript
import { relay } from '@relayplane/sdk'

// Configure providers and MCP servers
relay.configure({
  providers: {
    openai: { apiKey: process.env.OPENAI_API_KEY! }
  },
  mcp: {
    servers: {
      demo: { url: 'https://mcp-demo.relayplane.com' }
    }
  }
})

// Create workflow with AI + MCP steps
const workflow = relay
  .workflow('company-lookup')
  // Step 1: Extract company name (AI step)
  .step('extract')
  .with('openai:gpt-4o')
  .prompt('Extract the company name from this email: {{input.email}}. Return just the name.')
  // Step 2: Look up company (MCP step)
  .step('lookup')
  .mcp('demo:getCompanyInfo')
  .params({ companyName: '{{extract.output}}' })
  .depends('extract')
  // Step 3: Summarize (AI step)
  .step('summarize')
  .with('openai:gpt-4o')
  .depends('lookup')
  .prompt('Write a brief summary of this company: {{lookup.output}}')

// Run the workflow
const result = await workflow.run({
  email: 'Looking forward to our meeting! - John from Acme Corp'
})

if (result.success) {
  console.log('Company:', result.steps.extract)
  console.log('Info:', result.steps.lookup)
  console.log('Summary:', result.steps.summarize)
}
```

```json
{
  "success": true,
  "steps": {
    "extract": "Acme Corp",
    "lookup": {
      "name": "Acme Corp",
      "industry": "Technology",
      "employees": 500,
      "founded": 2015
    },
    "summarize": "Acme Corp is a technology company founded in 2015 with approximately 500 employees..."
  }
}
```

--------------------------------

### GET /v1/analytics/usage

Source: https://relayplane.com/docs/concepts/learning-ledger

Retrieves aggregate usage metrics from the ledger.

```APIDOC
## GET /v1/analytics/usage

### Description
Provides aggregate metrics such as total runs, total tokens, and total cost, grouped by specified dimensions.

### Method
GET

### Endpoint
/v1/analytics/usage

### Parameters
#### Query Parameters
- **workspace_id** (string) - Required - The unique identifier for the workspace.
- **group_by** (string) - Optional - Dimension to group metrics by (e.g., 'model').

### Response
#### Success Response (200)
- **total_runs** (integer) - Total count of runs.
- **total_tokens** (integer) - Total tokens consumed.
- **total_cost_usd** (number) - Total cost in USD.
- **breakdown** (array) - Detailed breakdown of metrics based on the group_by parameter.
```

--------------------------------

### Reference Previous Step Outputs in Prompts

Source: https://relayplane.com/docs/concepts

Use the `{{stepName.output}}` syntax to dynamically insert the output of a previous step into the system prompt of a subsequent step. Ensure dependencies are correctly defined.

```javascript
const workflow = relay
  .workflow('chained')
  .step('extract', {
    systemPrompt: 'Extract key points from the document'
  })
  .with('openai:gpt-4o')
  .step('summarize', {
    // Reference the extract step's output
    systemPrompt: `
      Summarize these key points:
      {{extract.output}}

      Keep it under 100 words.
    `
  })
  .with('anthropic:claude-sonnet-4-20250514')
  .depends('extract')
  .step('translate', {
    // Reference the summarize step's output
    systemPrompt: `
      Translate this summary to Spanish:
      {{summarize.output}}
    `
  })
  .with('openai:gpt-4o')
  .depends('summarize')
```

--------------------------------

### GET /v1/telemetry/health

Source: https://relayplane.com/docs/cost-optimization/usage-tracking

Retrieves the health status for each configured provider.

```APIDOC
## GET /v1/telemetry/health

### Description
Retrieves per-provider health status.

### Method
GET

### Endpoint
/v1/telemetry/health
```

--------------------------------

### Recommended Host Binding

Source: https://relayplane.com/docs/proxy/cli

For local-only access, bind the proxy to `127.0.0.1`. If network exposure is necessary, bind to `0.0.0.0` and ensure it's protected by a firewall, reverse proxy, and authentication.

```bash
# Recommended (local-only)
relayplane --host 127.0.0.1 --port 4100
```

```bash
# If you must bind all interfaces, put it behind firewall + reverse proxy + auth
relayplane --host 0.0.0.0 --port 4100
```

--------------------------------

### GET /v1/telemetry/savings

Source: https://relayplane.com/docs/cost-optimization/usage-tracking

Retrieves cost savings analysis data.

```APIDOC
## GET /v1/telemetry/savings

### Description
Retrieves cost savings analysis.

### Method
GET

### Endpoint
/v1/telemetry/savings
```

--------------------------------

### Upgrade RelayPlane Proxy

Source: https://relayplane.com/docs/proxy/cli

Check for proxy updates and install the latest version.

```bash
relayplane upgrade
```

--------------------------------

### Manage System Service

Source: https://relayplane.com/docs/proxy/cli

Installs or removes RelayPlane as a system service on Linux or macOS.

```bash
# Check service status
relayplane service status

# Install as system service (Linux: requires sudo)
sudo relayplane service install

# Preview what would be installed (dry run)
sudo relayplane service install --dry-run

# Remove system service
sudo relayplane service uninstall
```

--------------------------------

### Manage TLS with Certbot

Source: https://relayplane.com/docs/vps-deployment

Install Certbot and obtain a TLS certificate for the proxy domain.

```bash
# Install Certbot
sudo apt-get install -y certbot python3-certbot-nginx

# Obtain a certificate (nginx plugin handles nginx config automatically)
sudo certbot --nginx -d proxy.yourdomain.com

# Certbot adds a cron/systemd timer — verify auto-renewal works
sudo certbot renew --dry-run
```

--------------------------------

### Configure OpenAI Provider

Source: https://relayplane.com/docs/providers

Set up the OpenAI provider by providing your API key. This configuration should be done once at application startup.

```typescript
import { relay } from '@relayplane/sdk'

// Configure your provider (do this once at app startup)
relay.configure({
  providers: {
    openai: { apiKey: process.env.OPENAI_API_KEY! }
  }
})

const workflow = relay
  .workflow('openai-example')
  .step('analyze')
  .with('openai:gpt-5.2')
  .prompt('Analyze the provided text and extract key insights: {{input.text}}')

const result = await workflow.run({ text: 'Your content here...' })
```

--------------------------------

### Telemetry Payload Example

Source: https://relayplane.com/docs/telemetry

The structure of the anonymous metadata payload generated for every request.

```json
1{
2  "device_id": "anon_8f3a2b1c...",
3  "task_type": "code_review",
4  "model": "claude-sonnet-4",
5  "tokens_in": 1847,
6  "tokens_out": 423,
7  "latency_ms": 2341,
8  "success": true,
9  "cost_usd": 0.02
10}
```

--------------------------------

### Implement a Cost-Optimized Workflow

Source: https://relayplane.com/docs/concepts

Shows how to mix high-capability and low-cost models across sequential steps to reduce total expenditure.

```javascript
1const costOptimizedWorkflow = relay
2  .workflow('cost-optimized')
3
4  // Complex reasoning: Use most capable model
5  .step('analyze', {
6    systemPrompt: 'Perform deep analysis of this legal contract...',
7    maxTokens: 2000
8  })
9  .with('anthropic:claude-sonnet-4-20250514')  // $3/1M input tokens
10
11  // Simple extraction: Use cheaper model
12  .step('extractDates', {
13    systemPrompt: 'Extract all dates from: {{analyze.output}}'
14  })
15  .with('openai:gpt-4o-mini')  // $0.15/1M input tokens
16  .depends('analyze')
17
18  // Formatting: Use fastest/cheapest
19  .step('format', {
20    systemPrompt: 'Format as JSON: {{extractDates.output}}'
21  })
22  .with('google:gemini-2.0-flash')  // Very cost-effective
23  .depends('extractDates')
24
25// Result: 10x cost reduction vs using Claude for everything
```

--------------------------------

### GET /v1/runs/{run_id}/events

Source: https://relayplane.com/docs/concepts/learning-ledger

Retrieves the detailed event stream for a specific run.

```APIDOC
## GET /v1/runs/{run_id}/events

### Description
Fetches the detailed event stream for a specific run, useful for debugging and auditing the lifecycle of an AI operation.

### Method
GET

### Endpoint
/v1/runs/{run_id}/events

### Parameters
#### Path Parameters
- **run_id** (string) - Required - The unique identifier of the run.

### Response
#### Success Response (200)
- **events** (array) - A list of event objects containing event_id, event_type, timestamp, and data.
```

--------------------------------

### GET /v1/runs

Source: https://relayplane.com/docs/concepts/learning-ledger

Retrieves a list of recent runs filtered by workspace and optional parameters.

```APIDOC
## GET /v1/runs

### Description
Retrieves a list of recent runs for a specific workspace, with support for filtering by authentication type, status, and time range.

### Method
GET

### Endpoint
/v1/runs

### Parameters
#### Query Parameters
- **workspace_id** (string) - Required - The unique identifier for the workspace.
- **limit** (integer) - Optional - Maximum number of runs to return.
- **auth_type** (string) - Optional - Filter by 'api' or 'consumer'.
- **status** (string) - Optional - Filter by 'pending', 'running', 'completed', 'failed', or 'blocked'.
- **since** (string) - Optional - ISO 8601 timestamp to filter runs created after this time.
```
Tokens:
Raw