Spaces & Memory

Spaces are where your work lives. Memory is how agents learn about your project over time.

What is a space?

A space is a workspace container. Each space has:

• Tasks — with their own numbering sequence (EXP-1, EXP-2, etc.)
• Agents — configurations available within this space
• Memory — persistent context that agents can access
• Files — reference materials connected to the space
• Labels — custom tags for categorizing tasks

You might organize spaces by:

• Project — one space per product or codebase
• Client — one space per client or engagement
• Domain — one space per area of work (engineering, marketing, operations)

Creating a space

1. Click + New Space in the sidebar.
2. Give it a name, icon, and color.
3. Optionally, point it to a directory on your machine — this gives agents access to that codebase.
4. Set a task prefix (e.g., "EXP") for auto-incrementing task numbers.

What is memory?

Memory is persistent context that lives at the space level. Anything you add to memory is available to every agent working on tasks in that space.

Think of it as the "team knowledge base" — the things a new teammate would need to know when joining the project.

What to put in memory

Good memory entries include:

• Technical decisions — "We use Zod for validation", "API follows REST conventions"
• Project context — "This is a B2B SaaS product for healthcare", "Target audience is enterprise developers"
• Coding standards — "TypeScript strict mode", "Prefer composition over inheritance"
• Architecture notes — "Frontend is Next.js with app router", "Database is PostgreSQL with Prisma"
• Preferences — "Use Tailwind classes, not inline styles", "Write tests for all new functions"

What NOT to put in memory

• Temporary information — use the task description instead
• Task-specific context — put it in the task, not in memory
• Entire documents — memory should be concise, not a knowledge dump

How agents use memory

When an agent starts a task, it automatically receives relevant memory entries as part of its context. This means:

• You don't have to repeat project conventions in every task description
• New agents in the same space immediately understand the project
• Memory accumulates over time as you add more context

Adding memory

1. Open a space and go to the Memory section.
2. Click Add Memory.
3. Write your context entry — keep it clear and concise.
4. Save.

The memory is immediately available to all agents in that space.

Multi-space workflows

For larger projects, you might use multiple spaces:

Each space has its own memory, so agents working on docs aren't distracted by engineering standards, and vice versa. But you can use the same agent across spaces — it adapts to each space's context.

Space settings

Each space can be configured with:

• Name, icon, color — for visual identification
• Path — directory on your machine (for code projects)
• Prefix — task numbering prefix (e.g., "EXP")
• Memory enabled — toggle persistent memory on/off
• Labels — custom color-coded tags for task categorization