Coresearch
Use Coresearch for ongoing research work organized around a project and its conversations. It has its own project files and research surfaces; a study is the separate place to follow stage-scoped work and required review. You need a workspace and a healthy supported runtime on your own account to start execution.
Start and return to work
- Open New and choose the research flow that matches your task.
- Describe the question or material you want to explore. Select an available CLI and model before starting.
- Open Coresearch to return to projects you started. The list shows the project, CLI, lead and update time.
- Open the project and choose the conversation you want to continue.
For an existing conversation, execution settings are retained. If a desired runtime is absent from the picker, check your own online, healthy runtimes in Machines.
Inspect the project surfaces
In a research project, open the experiment surfaces to inspect Notebook, Files and Compute. Use Files to find project material and Notebook to inspect recorded computation. Read the conversation alongside the artifacts it references. Writing projects use their own writing surface.
When editing a file, pay attention to unsaved changes before switching material. Open the referenced file version when inspecting a transcript artifact so you know which content the conversation used.
Keep work understandable
Use a clear project name and separate conversations when the question changes. The Coresearch list supports renaming and archiving projects; review the archive confirmation before applying it. Returning to an existing project preserves its context better than starting duplicate work.
Success means you can reopen the project, continue the intended conversation and locate the material it produced. If no execution starts, inspect the runtime and specific error. If you need a staged research process, create a study; use Aristotle for questions about workspace evidence.