Can this hold the rationale behind design decisions, not just the tokens?

Yes, that is the point. Agents record durable observations like why a pattern was rejected or which spacing scale is canonical, and read them back so the reasoning, not just the values, guides future work.

Does it work across both design and engineering tools?

Yes. Glen is a standard MCP server, so the same memory is readable from design-to-code agents and engineering agents alike, keeping both sides aligned on the system.

Shared memory for designers

As a designer you work in a web of decisions, design tokens, component variants, accessibility rules, the reasons a past pattern was rejected, and increasingly you and your engineers use AI agents to translate designs into code, audit interfaces, and generate variations. Those agents have no memory of your design system or the rationale behind it, so they reinvent spacing scales, miss your accessibility standards, and repeat patterns your team already decided against. Glen, shared memory for AI agents, gives every agent your team uses one shared, durable memory exposed as a single MCP tool, so your design decisions persist and are recalled instead of relearned.

Design quality is consistency over time, and consistency is exactly what AI agents erode when they have no shared memory of your system. An agent generating a component does not know your team standardized on an 8-point spacing grid, that you deprecated the old button variant, or that a particular color combination fails your contrast bar, so it produces work that drifts from the system and lands review comments that someone has to write yet again. Glen turns those repeated corrections into durable memory: connected over MCP, every agent reads the shared store before it generates or audits, and writes back the decisions it is told, so the next generation respects them automatically.

For designers this means your design system finally has a memory that the tools doing the work actually consult. The rationale behind a pattern, why you chose this layout, why that interaction was rejected, lives next to the tokens themselves and is read at the moment an agent is about to repeat a past mistake. Because Glen is org-scoped, the memory is shared across your whole design and engineering organization, so a decision made in one project guides agent work in another, and a new designer's agents inherit the accumulated taste of the team. As a standard MCP server, Glen sits alongside the design and design-to-code tools your team already uses, complementing them with the why that lives in no Figma file. Connect once over OAuth or an API key and let your design decisions compound across every agent rather than evaporating after each handoff.

FAQ

Can this hold the rationale behind design decisions, not just the tokens?: Yes, that is the point. Agents record durable observations like why a pattern was rejected or which spacing scale is canonical, and read them back so the reasoning, not just the values, guides future work.
Does it work across both design and engineering tools?: Yes. Glen is a standard MCP server, so the same memory is readable from design-to-code agents and engineering agents alike, keeping both sides aligned on the system.