SpecWave

When juggling Cursor, Claude Code, Copilot, Gemini CLI and other AI coding assistants simultaneously, coding standards tend to scatter across per-platform config files that are expensive to maintain and easy to let drift. SpecWave solves this with a single declarative adapt table that natively lands unified coding standards, Harness gate processes, and IDE commands across 13 host environments, eliminating the need to hand-write rule files for each platform.

Core capability revolves around the Harness process: two file-level objects, task.md and spec.md, drive P0 gate verification, human-gate audit tables, and workflow orchestration. Cursor receives kit-* rules and command palette entries, Claude Code gains /kit: slash commands, and DSH exposes discoverable orchestration skills. Coding standards are injected into runtime context on demand via a tool call rather than rewritten at install time, so control stays with the developer.

Built for teams and individual developers who switch between multiple AI coding assistants and need a single discipline: if you use two or more hosts, do not want to maintain a separate set of coding standards per platform, and want CI pipelines to reuse the same gate logic, SpecWave's single npm package with a declarative adapt table is purpose-built for that.