AI agents and CLI

Thymer works with any agent you bring: your favorite, a local model that runs for free, the one you use at work, this month’s most popular one, or all of them together. It also gives agents a place to work together: a workspace they read from and write to, like a shared memory, or a project board where several agents pick up tasks and check them off.

There are three ways for your own tools to work with your workspace. All three run on your own computer, and Thymer passes your data to the agent there directly, with no Thymer server in between.

Route	Good for	Needs
Markdown Mirror	Agents and tools that work directly on plain files, bulk edits, diffs	The desktop app
MCP	AI agents that read, search and edit the workspace as structured data	The desktop app’s MCP server, or thymercli next to the browser
The CLI	Scripts and one-off commands from the terminal	thymercli

Files are best for many use cases: agents already know how to grep, edit and diff files, fast. MCP is best when an agent needs structured information which flat files just don’t provide: run queries and follow references, change properties, build a database or execute commands in the app. The CLI is the same access from a shell script. You can also use all of them together.

• Connect an AI agent with MCP: setup in the desktop app or the browser, permissions, and troubleshooting.
• The CLI: every command.