evidentloop/sopify

What it solves

Sopify addresses the issue of "premature coding" in AI development, where AI tools start writing code before requirements are fully understood or decisions are finalized. It prevents the loss of context when switching between different AI hosts or teammates and stops important design trade-offs from disappearing into ephemeral chat histories.

How it works

Sopify acts as a protocol layer that integrates into existing AI coding hosts (such as Codex, Claude, Cursor, Copilot, and Qoder). It enforces a workflow where the AI must plan first—analyzing, designing, and splitting tasks—before executing code. These plans, along with verification receipts and decision traces, are stored as git-tracked files in a .sopify/ directory within the repository. This allows any supported host to resume a task from a specific checkpoint using the ~go command, ensuring a consistent state across different tools and collaborators.

Who it’s for

Developers who use AI coding assistants and want a more structured, auditable, and resumable development process that persists across different AI tools and team members.

Highlights

• Plan-First Workflow: Forces the AI to stop and ask for clarification or approval before writing code.
• Cross-Host Resumability: Allows a developer to start a task in one AI tool and finish it in another without re-explaining the context.
• Git-Tracked Traceability: Stores plans and decisions as project artifacts in the repo, making the development history auditable.
• Host Agnostic: Works as an installation into existing editors and CLIs rather than requiring a new standalone tool.