A great data engine is completely useless if the developer experience (DX) makes you want to tear your hair out.
When building tools for AI agents, DX is twice as difficult. You aren't just designing a clean API for the human engineer writing the code; you are also designing an intuitive interface for the AI agent that will be manipulating the system autonomously.
In this post, we’re looking at how to build with thingd using the local CLI and our native TypeScript SDK to create low-latency, local-first agent loops.
Local-First DX: The thingd CLI
When an agent is autonomously modifying data, updating system states, or resolving something complex like our npm lookup tool on engine.thingd.cloud, human developers need a window into what the hell is happening under the hood.
The thingd CLI acts as that bridge. It allows you to spin up the local engine, inspect the object memory pools, and view real-time logs of what your agents are querying.
bash
# Start the local engine instance
thingd start
# Inspect live agent-created object entities
thingd objects:list --type memory_stream
Top comments (0)