LangChain Open-Sources Design Agent for Slack and Chat Workflows

LangChain has open-sourced Content Studio, a design agent built on Managed Deep Agents that generates on-brand visual assets directly from chat threads. The repository is available on GitHub under the Apache 2.0 license.

Overview and Capabilities

• Chat-Driven Workflow: Users can interact with the agent in Slack or via other supported interfaces, requesting assets such as resized ads, presentation decks, charts, and motion graphics.
• Sandbox Environment: Each thread runs in its own LangSmith sandbox, utilizing tools like Chrome, FFmpeg, and HyperFrames to render and check code-based assets without exposing API keys to the sandbox.
• Modular Skills and Memory: The agent uses a router skill to select from dozens of workflows, while maintaining team and personal memory layers to retain brand rules and feedback.

Deployment and Verification

According to the documentation, deploying the agent requires a LangSmith account—where Managed Deep Agents is currently in public beta on LangSmith Cloud—along with a compatible model key, such as Claude Opus 5.5 or endpoints supported via the LangSmith Gateway.

• Local testing can be performed using Docker with a scripted model and no API keys.
• Users can install dependencies using tools like `uv` and execute deployment commands as outlined in the repository instructions.
• The open-source code is free under the Apache 2.0 license, though users incur standard costs for model providers and Managed Deep Agents hosting.