Upsolve-Labs

Upsolve-Labs / upstack

Public

Lightweight Red/Green TDD skill suite for Claude Code, Cursor, Codex, and Gemini CLI. Tests before code, always.

upsolve.ai ai-coding claude-code developer-tools skills tdd
10
0
100% credibility
Found Mar 24, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Shell
AI Summary

upstack provides a suite of skills for AI coding assistants that guide users through a structured test-driven process for planning, implementing, validating, reviewing, and shipping code changes.

How It Works

1
🔍 Discover upstack

You hear about this handy toolkit that supercharges your AI coding buddy to build reliable features step by step.

2
⚙️ Quick setup

Run a simple installer that adds these special abilities to your AI helper in just 30 seconds.

3
💻 Open your project

Load up your code project in your AI coding tool and ask for a quick status check.

4
🗺️ Plan the work

Your AI reviews everything, suggests the best approach, draws up tests first, and breaks it into easy tasks with a clear roadmap.

5
🛠️ Build and test

Your AI writes failing tests first, then makes them pass by adding the code, keeping everything focused and reliable.

6
🔍 Check and review

Your AI explores the app, takes screenshots of how it works, spots any issues, and gives a thorough thumbs-up.

🚀 Ship it live

Your AI prepares a complete update with proof pictures and shares it ready for your team to approve and merge.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is upstack?

upstack is a lightweight Shell-based skill suite for AI coding tools like Claude Code, Cursor, Codex, and Gemini CLI, enforcing a Red/Green TDD workflow where tests come before code, always. It delivers slash commands such as /plan for scoping features with test proposals and dependency graphs, /execute for writing failing tests then implementations, /validate for browser-based verification with screenshots, /review for bug hunting, and /ship-pr for evidence-packed PRs. Built by Upsolve's CTO for reliable data agents, it solves flaky AI-generated code by grounding outputs in auditable tests and real browser evidence using tools like GitHub CLI and Vercel's agent-browser.

Why is it gaining traction?

Unlike heavier stacks, this github lightweight alternative clocks in at under 500 lines with zero bloat—no build systems or binaries—for 30-second installs via a smart shell script that handles OS-specific deps. Developers hook into its rigid plan-execute-validate-review-ship flow, which auto-generates Linear tickets or TODOS.md, uses agent-browser for UI traversal, and ensures scope discipline by parking discoveries. It's a focused evolution of gstack, tailored for iterating on existing projects at upstack company scale, appealing to those chasing upstack careers in secure AI engineering.

Who should use this?

AI-augmented backend devs building SaaS telemetry or data agents, frontend teams validating UIs via browser automation, and upstack freelancers/upstackers technologies handling bug fixes or features in Claude/Cursor. Ideal for upstack inc hq teams or solo devs wanting TDD guardrails before diving into lightweight github client workflows, especially with Linear integration.

Verdict

Try it for small AI-assisted sprints—solid docs and workflow make the 10 stars and 1.0% credibility score forgivable for such an early, lightweight github tags project. Skip for production unless you're okay auditing the minimal shell logic yourself.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.