Feature request: Let teammates’ AI agents collaborate around shared Figma assets
Our team uses Codex with Figma to build and maintain a shared component library. Multiple designers work with their own agents, each with different tasks and context. Developers may also have their own coding agents working on the corresponding implementation.
We can already see other collaborators’ agents and their working status in Figma. That made us wonder: could these agents collaborate with one another around shared design assets, with their owners’ permission?
The gap we encounter
An agent can inspect a shared component, but it may not know why another designer made certain decisions, which business constraints apply, or which questions remain unresolved. Today, we have to copy context and questions between separate conversations, relay the answers, and reconcile the results ourselves.
A workflow we would find useful
1. A designer’s agent updates a component and shares an approved brief describing the goal, changes, constraints, asset version, and open questions.
2. Another designer authorizes their agent to review that component against the shared design system.
3. A developer authorizes their coding agent to check the corresponding implementation and return discrepancies and supporting evidence.
4. The agents clarify differences in an asset-linked discussion and produce a summary of agreed items, required changes, and decisions that need a human.
5. The responsible people review and approve the outcome.
Suggested starting point
A lightweight “request a review from a teammate’s agent” feature could be a useful first step. Requests and replies would be attached to specific components or frames, reference the versions checked, and remain visible to the participating people.
Useful capabilities could include:
• Agent task cards showing ownership, asset scope, and availability for collaboration.
• Selective sharing of task summaries, constraints, and evidence.
• Structured review replies with findings and unresolved questions.
• Notifications when an asset changes after a review.
• An auditable history of requests, proposed changes, and human approvals.
Human control matters
This should be opt-in and stay within each person’s permissions. Sharing a Figma file should not automatically expose private agent conversations, repositories, or local files. Reading, replying, and editing should have separate permissions, with limits on automatic exchanges and cost. Final approval should remain with people.
We would love this to work across Codex and other compatible agent clients.
Is there already a supported workflow for this that we have missed? If not, would the Figma team consider asset-scoped collaboration between teammates’ agents?
Are other teams encountering the same context-sharing and review challenges?