Getting Started
Get a quick briefing on a codebase. Reads CLAUDE.md, recent git history, and audit reports to orient you.
Use when: Opening a project for the first time, or picking up after a break.
Deep investigation of a codebase. Produces an architecture map, quality scores, and actionable findings.
Use when: You inherited a project, or want to know what shape things are in.
Set up project governance — CLAUDE.md, AGENTS.md, rules, and hooks tailored to your project.
Use when: After an audit on existing projects, or when starting fresh.
Building Something New
Explore what you're building and why. Helps you think through the idea before jumping to code.
Use when: You have a vague idea. "I want to add a client portal" — start here.
Write a lightweight specification with clear acceptance criteria. Defines what "done" looks like.
Use when: You know what you want but need to nail down the details.
Create a step-by-step implementation plan with file lists, key decisions, and rationale.
Use when: You know what to build and need to figure out how.
Execute an implementation plan. Routes to the right execution strategy (sequential or parallel).
Use when: You have a plan and it's time to write code.
Fixing Something Broken
Systematic debugging. Investigates the root cause before proposing fixes — no guessing.
Use when: Something's broken and you don't know why.
Shipping
Emergency deploy. Lint, test, build, commit, push, deploy — minimum viable safety checks, maximum speed.
Use when: Something's broken in production and you fixed it. Get it live NOW.
Full due diligence. Code review, coverage audit, changelog, clean commits, PR, verification, deploy.
Use when: You finished a feature and want everything solid before it goes live.
Quality Checks
Multi-perspective review. For plans: engineering, CEO, or design review. For code: 3-pass analysis.
Use when: You want a second opinion on a plan or code before shipping.
Headless browser testing. Navigates your site, clicks things, finds bugs, takes screenshots.
Use when: You want to test your site the way a real user would.
Security audit using OWASP Top 10 and STRIDE threat modeling.
Use when: Before launching, or when you've added auth, payments, or user data handling.
Visual design audit. Finds spacing issues, hierarchy problems, and generic AI-generated design patterns.
Use when: Your site works but looks "off" and you can't pinpoint why.
Learning & Reflecting
Capture knowledge from non-obvious work. Creates a solution doc so you don't solve the same problem twice.
Use when: You just fixed something tricky and want to remember how.
Weekly engineering retrospective. Analyzes commits, work patterns, and code quality trends.
Use when: End of the week. What went well? What didn't?
Typical Flow
You don't need every command every time. Here's what a typical feature looks like:
Safety runs automatically. When you installed Soloship, it added hooks that scan for security issues, flag risky changes, and create rollback snapshots — all without you doing anything. The commands above are your workflow. The safety net is always on.
If Things Go Wrong
npx soloship rollback
Undo everything the AI did in the current session. Your pre-session work is preserved. Run this in your terminal (not in Claude Code).