clawplays/ospec

What it solves

AI coding assistants often store critical requirements and decision-making history only in chat logs, making it difficult to inspect, review, and maintain a consistent state across different sessions or different AI tools. OSpec provides a spec-driven development (SDD) layer that persists the change context—proposals, designs, plans, and verification evidence—directly within the project repository.

How it works

OSpec implements a "Loop Engineering" workflow (plan $\rightarrow$ act $\rightarrow$ verify) that integrates with AI agents like Claude Code, Codex, and Gemini. It operates through a CLI tool that manages the lifecycle of a requirement:

1. Initialization: Sets up a project structure (including a .ospec/ directory) to hold AI guidance and project knowledge.
2. Change Management: For simple tasks, it uses a fast flow (ospec change) that moves from proposal to tasks, implementation, and verification.
3. Goal Workflow: For complex work, ospec goal triggers a more rigorous agentic process involving design documents, task graphs, parallel worker dispatch, independent reviews, and automated verification gates.
4. Archiving: Once a change is validated and deployed, it is archived within the repo to maintain a permanent record of the work.

Who it’s for

Software developers who use AI coding agents and want a verifiable, repository-based audit trail of requirements and implementation plans rather than relying on ephemeral chat histories.

Highlights

• Spec-Driven Development: Turns chat requests into persistent files (proposals, designs, plans) stored in the repo.
• Agentic Controller: Supports parallel worker dispatch and reviewer gates for complex goals.
• Tool Agnostic: Works across various AI harnesses including Claude Code, Codex, Gemini, and OpenCode.
• Durable State: Saves progress in the repository so any assistant can resume a task exactly where the last one stopped.