What an agent can do
Once connected, an agent can:- Work on projects — list your projects, check one out as plain files, edit its blocks, parameters, subject, and variations, and commit the result back.
- Build and edit libraries — check out a whole library (its blocks, HTML, and parameters), make sweeping edits with real developer tools, and commit them as a new version.
- Check its own work — render a project or library preview, take screenshots of pages in a headless browser, and iterate until things look right.
- Find and create images — search your team’s image library, save new assets, generate images from a prompt, and upload files from its workspace.
- Read and write your docs — use your team’s knowledge base (brand guidelines, briefs, notes) as context, and keep it up to date.
- Send feedback — file bug reports or feature requests to the Denada team on your behalf.
How editing works
For projects and libraries, agents use a checkout → edit → commit model, much like version control:- The agent checks out the project or library as a folder of ordinary files (JSON and HTML).
- It edits those files locally with its own tools.
- It commits the whole thing back in one all-or-nothing step, with a short message describing the change.
Built-in guides
You don’t need to teach your agent how Denada works. The MCP server includes aread_guide tool with built-in documentation the agent discovers and reads on its own — guides covering libraries, the library authoring format, projects, images, email HTML compatibility, docs, and more. A well-behaved agent will consult these before making changes.
Where to start
The quickest entry point is inside Denada itself: in the chat sidebar of the library editor or project editor, look for Edit with external tools. It generates your API token and gives you a ready-to-paste connect command for your agent of choice. Or set things up manually — see Connecting your agent.Agents authenticate with your personal API token and can only see and change what you can. AI image generation additionally requires your team’s AI images setting to be enabled.