Skip to content

Keyboard

Shortcuts

Everyone learns Esc. These are the rest — all checked against Claude Code itself, not copied off a blog.

The three to learn first

If you remember nothing else.

  • Esc Esc
    Stop Claude mid-action The moment Claude starts doing something you did not intend.
  • Esc thenEsc Esc thenEsc
    Clear the input box Your half-written prompt has turned into a mess and you want a clean line.
  • Shift +Tab Shift +Tab
    Cycle permission mode Starting an exploration (plan) or letting an agreed plan run (accept edits). On older Windows setups without VT mode, this is Meta+M instead.

Writing the prompt

Composing, searching, pasting.

  • Ctrl +J Ctrl +J
    Start a new line without sending Any prompt longer than one sentence.
  • Ctrl +R Ctrl +R
    Search your prompt history "What did I type last week that worked?" Inside the search, Ctrl+S cycles the scope: this session → this project → everywhere.
  • Ctrl +S Ctrl +S
    Park the half-written prompt You are three sentences into a prompt and need to go check something.
  • Ctrl +G Ctrl +G
    Edit the prompt in your real editor Writing anything long or carefully structured.
  • Ctrl +V Alt +V
    Paste an image Anything visual. Annotate the screenshot first and it works even better. Ctrl+V, not Cmd+V — macOS keeps Cmd+V for its own paste. Windows and WSL use Alt+V. On WSL both Ctrl+V and Alt+V work.
  • Option +T Alt +T
    Toggle thinking for the next turn One hard question in the middle of otherwise simple work.
  • Ctrl +_ Ctrl +_
    Undo in the input box You deleted half a prompt by accident.

Seeing what happened

What Claude did and what it plans to do.

  • Ctrl +O Ctrl +O
    Toggle the full transcript You want to know what it actually did, not the summary.
  • Ctrl +T Ctrl +T
    Toggle the task checklist Any multi-step task, before it gets far.

Running the session

Long jobs, models, the terminal.

  • Ctrl +B Ctrl +B
    Send the running job to the background A long build, test run, or search you do not want to sit and watch.
  • Option +P Alt +P
    Switch model Between tasks — rarely a good idea mid-task.
  • Option +O Alt +O
    Toggle fast mode Straightforward work where you would rather have speed than depth.
  • Ctrl +Z Ctrl +Z
    Step out to the shell without quitting You need the terminal for one command and do not want to lose the session. Terminal-level shortcut (Unix SIGTSTP), not a Claude Code binding. Windows has no direct equivalent.

The keys are the same on both systems — only the modifier's name changes (macOS calls Alt Option). Run /keybindings in any session to see exactly what is bound on your machine.

Skills

Skills worth installing

A curated catalog of and — the ones we keep reaching for. Each entry tells you what the skill does, who it's for, the you type in Claude, and how to install it.

Show
Topic

NBG AI team · plugin library

NBG-AI/claude-tools — the team's developer skill library

An internal plugin library at NBG-AI/claude-tools — ~40 advanced skills for engineers actively working on NBG software.

New to Claude Code? Start at Foundations or browse the Tips and Use cases first.

1 — Access

TBD — the AI team is finalising the access-request process. For now, ping the AI team channel on Teams and they'll add you to the repo.

2 — Add the marketplace (once per machine)

/plugin marketplace add NBG-AI/claude-tools

3 — Install any skill from the library

/plugin install <name>@claude-tools

Built at NBG

Skills authored by the NBG AI team — tuned for our codebase, conventions, and compliance constraints.

  • Skill · Built at NBG

    build a TypeScript API sandbox from C# or Postman, end-to-end

    Type this in Claude /create-sandbox

    End-to-end pipeline for building a TypeScript implementation of an existing . Translate C# types and controllers to TypeScript (or infer types from a Postman collection), design the sandbox architecture, scaffold the multi-sandbox infrastructure, then implement service methods one at a time with integration + contract tests generated and run automatically.

    ~1-2 days of manual scaffolding per API advanced
    Use this when... you need a working TypeScript of an existing — for front-end work against a stable stub, integration testing, or contract validation. Starts from C# source or a Postman collection.

  • Skill · Built at NBG

    multi-agent build orchestrator

    Type this in Claude /team

    Orchestrates a multi- team — refiner, scanner, investigator, planner, designer, parallel coders, reviewer, dependency validator, test builder, integration verifier — to take a request from idea to merged code in one sitting.

    ~half a day per feature-sized request both
    Use this when... you have a feature-sized request — bigger than a one-line fix, smaller than a multi-week . It runs the full ten-phase pipeline (plan, design, code, review, test, verify) in one sitting.

  • Skill · Built at NBG

    deploy to NBG Azure

    Type this in Claude /deploy

    Automates end-to-end of NBG Azure-hosted web applications via the Azure Management Portal REST — first-deploy (provision), auth-setup (identity team credentials + proxy + Kafka), and redeploy (rebuild + push).

    ~45 min per deploy; ~half a day for a first-deploy advanced
    Use this when... shipping to NBG Azure — provisioning a brand-new app, wiring identity-team auth and Kafka audit, or pushing changes to an existing one. Skips the click-through portal grind.

  • Skill · Built at NBG

    User Acceptance Testing system

    Type this in Claude /uat-panel

    Conversational UAT — Claude walks a non-developer through validating a built feature, asks scripted questions, and records pass/fail evidence the team can sign off on.

    both
    Use this when... the app is ready for business UAT — Claude walks the reviewer through each acceptance criterion live and produces a clean pass/fail sign-off record.

Official — from Anthropic

Skills published and maintained by Anthropic themselves. The safest external bet: same vendor as Claude Code, and they ship in the marketplace that is configured out of the box.

  • Skill · Official — from Anthropic

    Word, Excel, PowerPoint and PDF files

    Type this in Claude /document-skills

    The four document that power Claude's own file creation — `docx`, `xlsx`, `pptx` and `pdf`. Claude stops handing you to reformat and produces the actual Office file, with working formulas, real headings, and page numbers where they belong.

    ~30-60 min per document beginner
    Use this when... the deliverable is an actual file — a Word memo, an Excel with live formulas, a PowerPoint deck, or a PDF you need to read, merge or fill in.

  • Skill · Official — from Anthropic

    write a skill of your own

    Type this in Claude /skill-creator

    's official for building skills — drafts one from a description, then helps you test and tune it with eval cases so you can tell whether it actually fires when it should, instead of guessing.

    ~a day per skill you'd otherwise hand-write advanced
    Use this when... you have explained the same procedure to Claude three times and it should have been written down once.

  • Skill · Official — from Anthropic

    distinctive UI generation

    Type this in Claude /frontend-design

    's official frontend-design — generates production-grade components and pages with distinctive visual choices, avoiding the generic "AI-slop" aesthetic that gives away machine-generated UIs.

    ~1-2 hours per page both
    Use this when... you want a frontend that looks intentional, not algorithmic — when Claude's default Tailwind-card-and-shadow output would give away that a human didn't sit at the keyboard.

From the community

Skills published by the wider Claude Code community — vetted by us and worth installing.

  • Skill · From the community

    get interrogated before anything gets built

    Type this in Claude /grill-me

    Turns Claude into an interviewer that works through your plan as a decision tree, one round of numbered questions at a time — each with its recommended answer — until nothing is left silently assumed. It refuses to start building until you confirm you agree.

    ~a day of rework per project both
    Use this when... you have a half-formed idea and you know the expensive mistakes are the assumptions you haven't noticed yet.

  • Skill · From the community

    end a session so the next one starts sharp

    Type this in Claude /handoff

    Compresses the live conversation into a handoff document a fresh can act on — written for the next , not as a summary for you. Plain , saved to a temp directory, and it names which the next session should reach for.

    ~20 min of re-explaining per restart both
    Use this when... a has got long or slow and the work is not finished — before you hit /clear, not after.

  • Skill · From the community

    a tutor that remembers where you got to

    Type this in Claude /teach

    Turns the current folder into a teaching workspace and Claude into a tutor that keeps your progress in files. Learning is treated as a multi- thing — you can stop, come back next week, and pick up where you left off instead of re-explaining what you already know.

    replaces a half-day course beginner
    Use this when... you want to genuinely learn a topic rather than get an answer — and you know it will take more than one sitting.

  • Skill · From the community

    real diagrams, not boxes with drop shadows

    Type this in Claude /diagram-design

    Roughly forty diagram types — flowcharts, swimlanes, org charts, Gantt, sequence, ER, Sankey, Wardley maps — produced as self-contained HTML with inline SVG. It can also learn your brand colours once and apply them to everything it draws afterwards.

    ~1-2 hours per diagram beginner
    Use this when... the thing you owe someone is a picture — a process flow, an org chart, a current-state architecture — and you don't have a drawing tool or the patience for one.

  • Skill · From the community

    a second opinion that hasn't seen the first

    Type this in Claude /doubt-driven-development

    Before a non-trivial decision stands, it spawns a fresh-context reviewer whose job is to disprove it rather than approve it. Catches the failure where a long quietly promotes an assumption into a fact nobody re-checked.

    ~a day per avoided bad merge advanced
    Use this when... a wrong answer is expensive and would be cheaper to catch now than to debug later — security logic, migrations, anything irreversible.

  • Skill · From the community

    throwaway code that answers one question

    Type this in Claude /prototype

    Builds deliberately disposable code to answer a specific design question — does this flow feel right, is this data workable, which of these three layouts wins. The output is a decision, not a codebase, and the code is meant to be deleted.

    ~a week of spec arguments both
    Use this when... you are stuck arguing about a design in the abstract and it would be quicker to look at two versions of it.

  • Skill · From the community

    make Claude cite the docs

    Type this in Claude /source-driven-development

    Forces every framework-specific decision to be checked against official documentation and cited, instead of recalled from training data. Training data goes stale and get deprecated; this makes Claude show you the source so you can check it yourself.

    ~2-4 hours per deprecated-API rabbit hole both
    Use this when... being subtly out of date would be expensive — a library you don't know well, or anything a colleague will copy across a .

  • Skill · From the community

    Get Shit Done framework family

    Type this in Claude /gsd-*

    A light-weight meta-, context-engineering and spec-driven development system for . A family of commands (`:new-`, `gsd:plan-phase`, `gsd:execute-phase`, `gsd:progress`, etc.) for managing multi-week projects with persistent `.planning/` state and per-phase artefacts.

    ~half a day per phase on multi-week projects advanced
    Use this when... Use this for multi-week where you'll pause, resume, and need an audit trail. Persistent `.planning/` folder with roadmap and per-phase artefacts. Overkill for one-sitting work — try `/team` instead.