scv

---

Languages: English · 한국어 · 日本語

Team collaboration plugin for Claude Code.

Quick Start

The only command you need to remember is /scv:help. It diagnoses your project's state and tells you what to do next. No flags to memorize, no docs to read first — the plugin walks you through every step.

# In any Claude Code session:

# 1. Install
/plugin marketplace add https://github.com/wookiya1364/scv-claude-code
/plugin install scv@scv-claude-code

# 2. /scv:help takes over from here.
#    On first run it offers to hydrate this project (one click).
/scv:help

That's it. You only need to remember /scv:help — it diagnoses your project and routes to the right command. Pick a starting line:

Situation	Command
Don't know what to do next	/scv:help
Have an idea but no materials yet	/scv:help "I want to add a refund button" (v0.9.0+)
Want to find past work in the archive	/scv:help "how did we handle refunds last quarter?" (v0.10.0+)

Platform prerequisites (1-time):

• macOS: brew install bash once — SCV scripts use bash 4+ features (declare -A). Scripts auto-escalate to /opt/homebrew/bin/bash after install, so the system bash 3.2 stays untouched.
• Linux / WSL: bash 4+ is the default — nothing to do.
• Windows native (PowerShell/cmd): unsupported; use WSL or Git Bash.
• All platforms: curl, git, jq, gh (or glab) are recommended. /scv:help flags any missing dependency on its first line.

---

5-Minute Walkthrough

Scenario: "Add a refund button to the checkout page"

Min	Step	What happens
1	Drop materials into scv/raw/	meeting notes + spec PDF (Jira URL inside)
2	/scv:promote	URL detected · slug + title asked · PLAN.md + TESTS.md + FEATURE_ARCHITECTURE.md written
3	/scv:work <slug>	Implements · runs Playwright e2e · captures .webm
4	Auto PR	PR opens with GIF preview · Mermaid diagrams · Jira link — all attached
5	Review → merge → archive	Reviewer confirms via GIF in 5 sec · merge moves plan to archive · joins /scv:regression

Stuck at any step? Run /scv:help — it picks up your project's current state and tells you what's next.

---

The Loop

Drop material → refine into a plan + tests → implement → archive. Every archived plan's tests join the accumulating regression suite that runs against every future change.