@voltagent/[email protected]

Patch Changes

• #149 0137a4e Thanks @VenomHare! - Added JSON schema support for REST API generateObject and streamObject functions. The system now accepts JSON schemas which are internally converted to Zod schemas for validation. This enables REST API usage where Zod schemas cannot be directly passed. #87

  Additional Changes:

  • Included the JSON schema from options.schema in the system message for the generateObject and streamObject functions in both anthropic-ai and groq-ai providers.
  • Enhanced schema handling to convert JSON schemas to Zod internally for seamless REST API compatibility.

• Updated dependencies [0137a4e, 4308b85]:

  • @voltagent/[email protected]

@voltagent/[email protected]

Patch Changes

• #149 0137a4e Thanks @VenomHare! - Added JSON schema support for REST API generateObject and streamObject functions. The system now accepts JSON schemas which are internally converted to Zod schemas for validation. This enables REST API usage where Zod schemas cannot be directly passed. #87

  Additional Changes:

  • Included the JSON schema from options.schema in the system message for the generateObject and streamObject functions in both anthropic-ai and groq-ai providers.
  • Enhanced schema handling to convert JSON schemas to Zod internally for seamless REST API compatibility.

• #151 4308b85 Thanks @process.env.POSTGRES_USER! - feat: Agent memory can now be stored in PostgreSQL database. This feature enables agents to persistently store conversation history in PostgreSQL. - #16

  Usage

  import { openai } from "@ai-sdk/openai";
  import { Agent, VoltAgent } from "@voltagent/core";
  import { PostgresStorage } from "@voltagent/postgres";
  import { VercelAIProvider } from "@voltagent/vercel-ai";
  
  // Configure PostgreSQL Memory Storage
  const memoryStorage = new PostgresStorage({
    // Read connection details from environment variables
    connection: {
      host: process.env.POSTGRES_HOST || "localhost",
      port: Number.parseInt(process.env.POSTGRES_PORT || "5432"),
      database: process.env.POSTGRES_DB || "voltagent",
   || "postgres",
      password: process.env.POSTGRES_PASSWORD || "password",
      ssl: process.env.POSTGRES_SSL === "true",
    },
  
    // Alternative: Use connection string
    // connection: process.env.DATABASE_URL || "postgresql://postgres:password@localhost:5432/voltagent",
  
    // Optional: Customize table names
    tablePrefix: "voltagent_memory",
  
    // Optional: Configure connection pool
    maxConnections: 10,
  
    // Optional: Set storage limit for messages
    storageLimit: 100,
  
    // Optional: Enable debug logging for development
    debug: process.env.NODE_ENV === "development",
  });
  
  // Create agent with PostgreSQL memory
  const agent = new Agent({
    name: "PostgreSQL Memory Agent",
    description: "A helpful assistant that remembers conversations using PostgreSQL.",
    llm: new VercelAIProvider(),
    model: openai("gpt-4o-mini"),
    memory: memoryStorage, // Use the configured PostgreSQL storage
  });

@voltagent/[email protected]

Patch Changes

• #149 0137a4e Thanks @VenomHare! - Added JSON schema support for REST API generateObject and streamObject functions. The system now accepts JSON schemas which are internally converted to Zod schemas for validation. This enables REST API usage where Zod schemas cannot be directly passed. #87

  Additional Changes:

  • Included the JSON schema from options.schema in the system message for the generateObject and streamObject functions in both anthropic-ai and groq-ai providers.
  • Enhanced schema handling to convert JSON schemas to Zod internally for seamless REST API compatibility.

• Updated dependencies [0137a4e, 4308b85]:

  • @voltagent/[email protected]

@voltagent/[email protected]

Patch Changes

• #160 03ed437 Thanks @omeraplak! - feat: add Vercel AI SDK observability exporter

  • Introduce new @voltagent/vercel-ai-exporter package for Vercel AI SDK integration
  • Provides OpenTelemetry exporter for VoltAgent observability
  • Enables comprehensive tracking of LLM operations and multi-agent workflows
  • Includes automatic telemetry collection and agent history management

• Updated dependencies [03ed437, 03ed437]:

  • @voltagent/[email protected]
  • @voltagent/[email protected]

@voltagent/[email protected]

Patch Changes

• #160 03ed437 Thanks @omeraplak! - feat: enhanced Supabase memory provider with better performance

  We've significantly improved the Supabase memory provider with better schema design and enhanced performance capabilities. The update includes database schema changes that require migration.

  Migration commands will appear in your terminal - follow those instructions to apply the database changes. If you experience any issues with the migration or memory operations, please reach out on Discord for assistance.

  What's Improved:

  • Better performance for memory operations and large datasets
  • Enhanced database schema with optimized indexing
  • Improved error handling and data validation
  • Better support for timeline events and metadata storage

  Migration Notes:

  • Migration commands will be displayed in your terminal
  • Follow the terminal instructions to update your database schema
  • Existing memory data will be preserved during the migration

• Updated dependencies [03ed437]:

  • @voltagent/[email protected]

@voltagent/[email protected]

Patch Changes

• #160 03ed437 Thanks @omeraplak! - feat: introduce new VoltAgent SDK package

  • Add new @voltagent/sdk package for client-side interactions with VoltAgent API
  • Includes VoltAgentClient for managing agents, conversations, and telemetry
  • Provides wrapper utilities for enhanced agent functionality
  • Supports TypeScript with complete type definitions

• Updated dependencies [03ed437]:

  • @voltagent/[email protected]

@voltagent/[email protected]

Patch Changes

• #160 03ed437 Thanks @omeraplak! - refactor: remove peer dependencies and update package configuration

  • Remove @voltagent/core peer dependency from Google AI and Groq AI packages
  • Clean up package.json formatting and configuration
  • Improve dependency management for better compatibility

• Updated dependencies [03ed437]:

  • @voltagent/[email protected]

@voltagent/[email protected]

Patch Changes

• #160 03ed437 Thanks @omeraplak! - refactor: remove peer dependencies and update package configuration

  • Remove @voltagent/core peer dependency from Google AI and Groq AI packages
  • Clean up package.json formatting and configuration
  • Improve dependency management for better compatibility

• Updated dependencies [03ed437]:

  • @voltagent/[email protected]

@voltagent/[email protected]

Patch Changes

• #160 03ed437 Thanks @omeraplak! - feat: improved event system architecture for better observability

  We've updated the event system architecture to improve observability capabilities. The system includes automatic migrations to maintain backward compatibility, though some events may not display perfectly due to the architectural changes. Overall functionality remains stable and most features work as expected.

  No action required - the system will automatically handle the migration process. If you encounter any issues, feel free to reach out on Discord for support.

  What's Changed:

  • Enhanced event system for better observability and monitoring
  • Automatic database migrations for seamless upgrades
  • Improved agent history tracking and management

  Migration Notes:

  • Backward compatibility is maintained through automatic migrations
  • Some legacy events may display differently but core functionality is preserved
  • No manual intervention needed - migrations run automatically

  Note:
  Some events may not display perfectly due to architecture changes, but the system will automatically migrate and most functionality will work as expected.

@voltagent/[email protected]

Patch Changes

• 85204e2 Thanks @omeraplak! - feat: add provider options support including thinkingConfig - #138

  const response = await agent.generateText("Write a creative story.", {
    provider: {
      thinkingConfig: {
        thinkingBudget: 0,
      },
    },
  });