Practical overview
When it fits
Use Daydreams when the main requirement is experimenting with composable TypeScript agent architecture, MCP tool access or the repository’s agentic-commerce direction. Review the project notice first because the framework is no longer the core focus and some features are described as obsolete.
Who it is for
- TypeScript developers building or studying AI agents
- Teams prototyping commerce-oriented agent workflows
- Developers integrating MCP servers with agent actions
- Engineers evaluating context-composition patterns
Documented uses
- Assemble an agent from reusable contexts with their own state, instructions and actions.
- Connect an agent to external MCP resources and tools.
- Prototype commerce-oriented agent applications.
- Access multiple model providers through the documented Dreams Router interface.
Requirements & limits
The repository is public and lists an MIT license. The framework targets cloud, self-hosted and server environments in the technical project classification. The documented Router example includes USDC micropayments through x402, with a Base Sepolia example, but this does not establish general service pricing or availability.
- TypeScript is the documented primary language.
- Context composition is implemented through reusable contexts and .use().
- MCP connectivity is shown through an extension using configured stdio transports.
- The README says the project’s current focus is agentic commerce rather than the agent harness.
- The README recommends lucid-agents with the Pi agent harness for building agents.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Compose a support agent with analytics context
Create an agent whose support context includes analytics actions and conditional premium behavior.
Before you start
- TypeScript project
- Daydreams core package
- An AI SDK model provider
- A support context, analytics context and premium context as shown in the repository example
Steps
Define an analytics context with state for events and sessions.
Add a trackEvent action that records an event in context memory.
Define a support context with customer arguments and use() to include analytics, adding premium context conditionally.
Create the agent with the model and support context, then send an input with customer arguments.
Expected outcome
The agent receives support behavior together with analytics functionality and conditional premium features in one composed conversation.
Guide limitations
- The repository notice says the framework is no longer the core focus and that some features are obsolete.
- The example does not document a complete project setup or deployment procedure.
Connect an agent to MCP filesystem and database tools
Configure an agent extension that exposes filesystem resources and SQLite database tools through MCP.
Before you start
- TypeScript project
- Daydreams MCP package
- The MCP filesystem and SQLite servers used in the example
- A local docs directory and SQLite database
Steps
Create an MCP extension with stdio transports for the filesystem and SQLite servers.
Add the extension to the agent configuration.
Define an action that calls mcp.listResources for the filesystem server.
Call mcp.callTool for the database server from an agent action.
Expected outcome
Agent actions can request filesystem resources and invoke the configured database tool through MCP.
Guide limitations
- The example documents configuration and calls but does not establish a complete search implementation or deployment process.
- The database query shown in the repository example interpolates the query into SQL and should not be treated as a security recommendation.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 3 October 2026
- Original website title
- daydreamsai/daydreams · Official homepage title · retained in its source language