Spec Driven Development for Developers and AI Agents.
Turn endless iterations into one-shot features.
Shotgun writes codebase-aware specs for AI coding agents. We index your entire repo, research how features should fit your architecture, and generate technical specs that keep Cursor, Claude Code, and Lovable on track.
Export to AGENTS.md. Your AI agent builds correctly the first time instead of derailing halfway through.
Open source CLI for developers working with AI coding agents.
Feedback Wizard AI
Noice! Finally, something that will help me save tokens when building 🚀
I'd love to see a feature where I can pick a template that I can stay in. For example follow a design sprint methodology.
Shotgun CLI
@wozniakwo_ Glad you like it, Wojtek! 🤠 Templates are actually on our roadmap — exactly for flows like design sprints (and others, e.g. PRDs, trade-off docs, architecture patterns). The idea is you’ll be able to pick a template and let Shotgun guide you through it, while still staying flexible.
Curious: what would your “must-have” sections be in a design sprint template? 🚀
Shotgun CLI
@wozniakwo_ it's coming soon!
Tech To The Rescue
Shotgun CLI
@tomik99 Thanks so much, Tomasz 🙌 You put it perfectly
Your Next Store
Love this all-in-one approach - feels like the missing abstraction layer for building with AI.
Shotgun CLI
@zaiste Thanks Jakub 🙌 That “missing abstraction layer” phrasing really nails it — that’s exactly how we’ve been thinking about Shotgun. Coding agents are powerful, but without a spec layer in between it often turns into trial-and-error.
Enter the era of vibe shaping ✨⛏️ Gave it a go a few weeks back and was impressed with the depth of the research. Cool sparring partner to refine all angles of the `what` and `how`
Shotgun CLI
@k15z Appreciate you, Kacper 🙌 Love that “vibe shaping” take!!
Really interesting approach. I’m curious how well does Shotgun handle large, messy repos with legacy code? And when exporting specs to Cursor or Claude Code, does it preserve repo-specific context or do devs need to reconfigure things on the other side? Excited to see how this changes the workflow for teams that usually spend hours writing specs by hand.
Shotgun CLI
@eziyi_ruth Great questions, Ezeyi 🙏 Right now Shotgun can handle repos by indexing the structure and grounding specs in what’s actually there — though we’re still refining performance for very big codebases, so that’s active work during Alpha.
When exporting, the goal is to keep repo-specific context intact so you don’t have to reconfigure things on the other side. Cursor/Claude get a spec that’s already shaped to your repo, rather than something generic.
Team workflows are exactly where we see this making the biggest impact — replacing hours of manual spec-writing with structured artifacts you can iterate on. In your own projects, do you see the bigger bottleneck in understanding messy existing code, or in spec creation from scratch?
Shotgun CLI
@irene_morrison1 We only gather minimal, anonymous events (e.g., install, server start, tool call). We don’t collect the content itself—only that an event occurred. We use Sentry for error reporting to improve stability.
We’ve just released the beta version. Up for testing?https://github.com/shotgun-sh/shotgun-alpha
Hi folks, I’m asking the same question that hasn’t been answered yet: how did your tool’s idea and name end up mirroring another tool that has existed longer? We’re waiting for your comments! https://github.com/glebkudr/shotgun_code/