From token-wise
Generate or update CLAUDE.md using Sonnet (balanced quality and cost). Use when the user selects Sonnet for CLAUDE.md tasks. — First time only: select "Yes, and don't ask again" to skip this approval prompt in future.
How this skill is triggered — by the user, by Claude, or both
Slash command
/token-wise:claude-md-sonnetWhen to use
Invoked via /token-wise:claude-md-sonnet.
claude-sonnet-4-6This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are running as **claude-sonnet-4-6** to save tokens on this text-heavy task.
You are running as claude-sonnet-4-6 to save tokens on this text-heavy task.
Run these commands to understand the project before writing anything:
![ -f CLAUDE.md ] && echo "EXISTING_CLAUDE_MD_FOUND" || echo "NO_EXISTING_FILE"
!ls -1 2>/dev/null | head -40
![ -f package.json ] && cat package.json | head -40 || true
![ -f composer.json ] && cat composer.json | head -40 || true
![ -f Cargo.toml ] && cat Cargo.toml | head -30 || true
![ -f pyproject.toml ] && cat pyproject.toml | head -30 || true
![ -f go.mod ] && cat go.mod | head -20 || true
![ -f .env.example ] && cat .env.example | head -30 || true
![ -f .env.local.example ] && cat .env.local.example 2>/dev/null | head -30 || true
![ -d .git ] && git log --oneline -5 2>/dev/null || true
![ -d .git ] && git remote -v 2>/dev/null | head -4 || true
!find . -name "*.info.yml" -path "*/modules/custom/*" 2>/dev/null | head -5 | xargs grep -l "type: module" 2>/dev/null || true
!find . -name "*.info.yml" -path "*/themes/custom/*" 2>/dev/null | head -3 || true
If EXISTING_CLAUDE_MD_FOUND was printed above, read the current CLAUDE.md and treat it as a base to update rather than replace entirely. Preserve any manual sections the user may have written.
Write a CLAUDE.md file to the project root. Follow these rules strictly:
Target: under 6,000 tokens total. Every section must earn its place.
Use this structure — include a section only if there is real content to put in it:
# Project: <name>
## Overview
One paragraph. What the project does, who it is for, tech stack summary.
## Architecture
Key directories and what lives in them. Only non-obvious structure needs explanation.
## Development setup
Commands to install, configure, and run. Prerequisites if non-standard.
## Common tasks
The most frequent dev tasks as short numbered steps or commands. No fluff.
## Code conventions
Language/framework-specific rules that apply here. Naming, patterns to follow or avoid.
## Testing
How to run tests. Where tests live. Any known gotchas.
## Environment variables
List required env vars with a one-line description each. No values.
## Important constraints
Things Claude must never do in this project (e.g. never edit generated files, never commit secrets, always run linter before proposing a PR).
Rules for content:
Write the final content to CLAUDE.md in the current working directory. Overwrite if it already exists.
After writing, print:
✅ CLAUDE.md written to <path>
Model used : claude-sonnet-4-6 (token-wise)
File size : <size>
npx claudepluginhub vivek-cactus/token-wise --plugin token-wiseGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.