While working on projects in Cursor, it’s often necessary to use multiple specialized chats at the same time — for example, for code generation, debugging, specific Git/GitHub operations, and other service tasks.
For each such chat, we need separate, individual rules and system prompts so that the AI agent can work as efficiently as possible in the given context.
I propose adding the ability, directly within the chat interface, to:
- Enable or disable project and user rules individually for that chat, rather than only via configuration/settings that affect all chats at once.
- Set a unique system prompt for each chat to flexibly adapt the agent’s behavior to the specific task or workflow.
- Easily switch between different rule sets (project and user) for quick experimentation, adapting to new tasks, or optimizing collaboration in a team.
This feature would make it possible to:
- Organize work on complex, multi‑context projects where AI rules and behavior differ for frontend, backend, documentation, tests, etc.;
- Increase flexibility, control, and quality in working with the AI assistant;
- Allow experienced developers and teams to configure their workspace for their own vibe‑coding and automation scenarios;
- Save tokens by using only the rules specific to each chat.
I believe this feature would be useful for many who actively use Cursor and AI agents in development!