tim-hub/get-started

The base directory for this skill is shown in the header when invoked (e.g. /path/to/plugin/skills/get-started). All script paths below use <base-dir>/scripts/ — substitute the actual base directory path shown in the header.

Arguments

This skill accepts optional arguments to run specific steps. With no arguments or all, every step runs.

| Argument | Step(s) | |----------------|----------------------------------| | all | Run all steps (default) | | claude.md | Step 1 — Set up CLAUDE.md | | statusline | Step 2 — Set up statusline | | playwright | Step 3 — Install playwright-cli | | chub | Step 4 — Install chub | | rtk | Step 5 — Install rtk | | skills | Step 6 — Upsert external skills | | rules | Step 7 — Upsert rules | | settings | Step 8 — Upsert Claude settings | | plugins | Step 9 — Upsert plugins |

Multiple arguments can be passed space-separated, e.g. /get-started rules plugins settings.

Parse the arguments from the skill invocation. If no arguments or all, run every step. Otherwise, run only the matching steps in order.

---

Before starting, create the sentinel file so the Stop hook knows this skill ran:

touch /tmp/.get-started-running

Step 1: Set up CLAUDE.md

Argument: claude.md

Check if ~/.claude/CLAUDE.md already exists:

test -f ~/.claude/CLAUDE.md && echo "exists" || echo "missing"

• If missing: run the script using the base directory from this skill's header:

  bash "<base-dir>/scripts/init-claude-md.sh"
  

• If exists: use AskUserQuestion to ask:

  "~/.claude/CLAUDE.md already exists. Override it with the powerball template?"

  • Options: "Yes, override" / "No, skip"
  • If Yes: run the script
  • If No: report Skipped: ~/.claude/CLAUDE.md

Step 2: Set up statusline

Argument: statusline

Check if ~/.claude/statusline-command.sh already exists:

test -f ~/.claude/statusline-command.sh && echo "exists" || echo "missing"

• If missing: copy the statusline script:

  bash "<base-dir>/scripts/init-statusline.sh"
  

  Then use the statusline-setup Agent to configure ~/.claude/settings.json with the command bash ~/.claude/statusline-command.sh.

• If exists: use AskUserQuestion to ask:

  "~/.claude/statusline-command.sh already exists. Override it with the powerball statusline?"

  • Options: "Yes, override" / "No, skip"
  • If Yes: run the script, then use the statusline-setup Agent to configure ~/.claude/settings.json
  • If No: report Skipped: ~/.claude/statusline-command.sh

Step 3: Install playwright-cli skills

Argument: playwright

Check if @playwright/cli is already installed globally:

playwright-cli --version 2>/dev/null && echo "installed" || echo "missing"

• If installed: skip the install step, proceed directly to running:

  cd ~ && playwright-cli install --skills
  

• If missing: install it first, then install skills:

  pnpm install -g @playwright/cli@latest
  cd ~  && playwright-cli install --skills
  

Step 4: Install chub and get-api-docs skill

Argument: chub

Check if @aisuite/chub is already installed globally:

chub --version 2>/dev/null && echo "installed" || echo "missing"

• If missing: install it first:

  pnpm install -g @aisuite/chub@latest
  

Then add the get-api-docs skill globally (whether it was already installed or just installed):

mkdir -p ~/.claude/skills/get-api-docs && curl -fsSL https://raw.githubusercontent.com/andrewyng/context-hub/main/cli/skills/get-api-docs/SKILL.md -o ~/.claude/skills/get-api-docs/SKILL.md

Report whether chub was already installed or freshly installed, and confirm the skill was added to ~/.claude/skills/get-api-docs/SKILL.md.

Step 5: Install rtk

Argument: rtk

Invoke the rtk-setup skill to handle RTK installation and configuration.

Step 6: Upsert external skills

Argument: skills

bash "<base-dir>/scripts/upsert-skills.sh"

Step 7: Upsert rules

Argument: rules

bash "<base-dir>/scripts/upsert-rules.sh"

Step 8: Upsert Claude Code settings

Argument: settings

bash "<base-dir>/scripts/upsert-claude-settings.sh"

Step 9: Upsert plugins

Argument: plugins

bash "<base-dir>/scripts/upsert-plugins.sh"

Completion

Report the output of each step that was run, then stop.