langchain-ai/open-canvasArchived

Features

• Collaborative AI Agent Runtimes - Provides a server-side runtime where humans and AI agents collaboratively produce and edit documents and code in real time.
• Stateful Agent Orchestration - Provides a framework for managing complex, stateful agent workflows with durable execution and human-in-the-loop interventions.
• Agent Memory Personalization - Generates and stores reflections and memories from chat history to personalize subsequent AI interactions.
• Contextual Memory Namespaces - Stores user preferences and reflections in categorized namespaces to retrieve context across different conversational threads.
• AI Collaborative Authoring - Provides a shared interface for humans and LLMs to collaboratively create and edit documents and code.
• Iterative Generative Coding - Supports developing software with an LLM via direct editing, version tracking, and live rendering of generated code.
• LLM Application Development - Provides a framework for building AI agents with custom middleware for guardrails, routing, and tool policies.
• LLM API Integrations - Provides capabilities for connecting to external and local large language model providers via standardized APIs.
• Provider-Agnostic Model Interfaces - Implements abstraction layers that standardize inputs and outputs across multiple LLM providers to ensure portability.
• Conversation Thread Contexts - Maintains persistent conversation state and user memories across threads to tailor LLM responses.
• AI Co-authoring Workspaces - Provides a shared workspace for real-time co-authoring of documents and code between humans and large language models.
• Agent State Persistence - Implements a persistence layer to track session context, user memories, and historical snapshots across conversational threads.
• Middleware-Style Guardrail Pipelines - Processes LLM requests through sequential chains of pluggable middleware modules for guardrails and routing.
• Durable Execution Persistence - Persists the current state of agent runs to a database to allow for human-in-the-loop interruptions and resumes.
• Durable Workflow Execution Engines - Ensures fault-tolerant process completion by persisting execution state and supporting human interventions.
• Session Context Persistence - Tracks ongoing conversations by persisting message history and stateful data within a single thread.
• Agent Management Interfaces - Provides a graphical interface for monitoring and interacting with stateful AI agent workflows and execution pipelines.
• Live-Rendering Markdown Editors - Provides a unified view for writing and editing structured code and markdown with integrated live rendering.
• Middleware Workflow Composition - Enables the construction of agent harnesses through a pipeline of middleware for guardrails, retries, and tool policies.
• LLM Orchestration - Implements a system to integrate multiple model providers through standardized interfaces for chat and embeddings.
• Local Agent Servers - Provides a local server that exposes endpoints for runs, threads, and assistants with an integrated database.
• Markdown Editors - Provides a structured text editor with live rendering and version tracking for AI-generated markdown and code.
• User Preference Persistence - Stores facts, experiences, and rules in custom namespaces to recall specific data across different conversational threads.
• Collaborative Artifact Snapshots - Saves historical snapshots of documents and code to review or revert to previous versions.
• Document and Code Snapshots - Maintains a historical snapshot system for documents and code blocks to enable auditing and version reversion.

Star history