Skill · Open source

shipshape

Dual-level plan and code review for AI coding agents: a staff-engineer pass, then a principal-engineer pass. Works with Claude Code, OpenCode, Codex, Gemini, and Cursor.

Role ORIGINATOR Stars ★ 3 Language Shell Pushed 21 Apr 2026

View on GitHub → Full README on GitHub →

Open sourceThe short version

What it does

Dual-level plan and code review for AI coding agents: a staff-engineer pass, then a principal-engineer pass. Works with Claude Code, OpenCode, Codex, Gemini, and Cursor.

Install

git clone https://github.com/lucasyhzhu-debug/shipshape.git ~/.claude/skills/shipshape

Source

github.com/lucasyhzhu-debug/shipshape

Why use it

It finds the gaps in a plan while they are still cheap to fix, before anyone writes the code.

More of the toolchain

VISIBILITY Markdown

ai-seo

Works on getting cited by ChatGPT, Perplexity, Claude and Google AI Overviews rather than ranked. Covers the visibility audit, extractable structure, machine-readable files like llms.txt and /pricing.md, and what agents need to read a page at all.

Why use itTwo gates bound every run. It will not start without a resolving target and a query to measure against, and it will not end without writing the findings to a file first — audit runs have a habit of dissolving into a closed session with nothing saved.

pushed 11 Aug read

AUDIT Markdown

seo-audit

A technical and on-page SEO audit skill that refuses to audit a target that is not there. It opens with a preflight gate: before reading its own next section, it checks for site files in the working directory or a live URL that resolves, and aborts naming what is missing.

Why use itPoint audit tooling at an empty directory and it will hand you a confident report about nothing. This one stops instead, and ships a table of the excuses an agent uses to talk itself past the gate.

pushed 11 Aug read

INSTRUMENTATION JavaScript

claude-code-statusline

A four-line Claude Code status line: context-window usage, the 5h and 7d limit windows with pace markers, git state, and cache stats. It runs on pure Node with no dependencies.

Why use itIt answers the two questions you keep asking mid-session: how full is the context window, and how close am I to the limit.

pushed 05 Aug read

Where this shows up in the writing

essay 2026-07

At McKinsey I was usually the dumbest person in the room

read