Integrating GSD with Cursor in a Real-World Project

[Abdeltoto]

Hi,

I’m exploring how to leverage GSD (Get Shit Done) within Cursor to structure AI-assisted development more systematically.
For those already using it in production:

• What does a robust GSD + Cursor workflow look like in practice?
• How do you structure specs, tasks, and iterations to maximize LLM efficiency while minimizing context drift?
• Do you maintain a distinct GSD planning layer alongside the codebase, or integrate it directly into the repository structure?
• If introducing GSD mid-project, how would you retrofit it without slowing down delivery?
• Is there a recommended migration strategy (e.g., progressive feature adoption, spec backfilling, or parallel planning layer)?

Appreciate insights from teams applying this beyond experimental use.

[trek-e]

Cursor is fully supported since v1.29. Install with:

npx get-shit-done-cc --cursor --global

GSD installs as Cursor skills (skills/gsd-*/SKILL.md). The workflow is identical to Claude Code — /gsd:new-project, /gsd:plan-phase, etc.

For Cursor-specific tips, GSD auto-converts tool names (e.g., Bash→launch-process) and brand references during installation.

---

Generated by Claude Code