Anima Reference Architecture

Overview

This architecture separates design-source intake, authenticated code generation, deterministic post-processing, and reviewed delivery. It is intended for repeatable Figma-to-code pipelines where each run is bounded to approved files and nodes, produces inspectable artifacts, and can be stopped or rolled back without exposing design content or credentials.

Prerequisites

• Define the target framework, repository layout, supported Anima/Figma SDK versions, and the owner who approves generated changes. Pin dependencies and create a sandbox Figma file with synthetic components for pipeline tests.
• Obtain Figma and Anima credentials through the deployment secret manager, using least-privilege scopes and short-lived credentials where supported. Validate the configured Figma webhook passcode before accepting events; never commit, print, or place tokens in generated code, cache files, pull requests, or receipts.
• Establish allowlists for Figma file IDs, node IDs, webhook sources, output repositories, and branch names. Define retention and deletion rules for source snapshots, generated output, and logs before enabling automation.
• Prepare a dry-run mode, an artifact-diff gate, a staged canary environment, and a rollback reference to the last approved generated revision. Do not allow a webhook to publish directly to production.

System Architecture