Claude Code Drop-In Kit

39 tested files you drop into .claude/. Not a PDF. Not a course.

Everyone else sells a guide about Claude Code. This is the actual files, wired into your team's repos. One command installs them stand-alone, or I roll them out across your organisation myself.

$1,499 Enterprise rollout · up to 5 repos · done-for-you
Book Enterprise rollout — $1,499 →

Set the coffee count to 300 ($1,500) and put "ENTERPRISE" plus your company name and repo count in the message, then email fayazdevinbox@uberip.com so I can start.

$999 Team licence · your whole organisation · forever
Buy the Team licence — $999 →

Set the coffee count to 200 ($1,000) and put "TEAM" plus your company name in the message, then email fayazdevinbox@uberip.com.

$249 Setup Sprint · one repo · delivered in 3 working days
Book a Setup Sprint — $249 →

Set the coffee count to 50 ($250) and put "SPRINT" plus your repo URL in the message, then email fayazdevinbox@uberip.com so I can start.

Just want the files yourself?

$50 one-time · one developer · unlimited projects · forever

Paid through Fayaz's Buy Me a Coffee page (Stripe, already live). A coffee is $5 — set the number to 10, and put "CCDK" in the message so he knows what it was for.

Buy the full kit — $50 (10 coffees) →
Thank you.
⬇ claude-code-drop-in-kit.zip (60 KB, 57 files)

Unzip, then from your project root run bash install.sh (or bash install.sh --global for every project). Restart Claude Code — hooks are only read at startup.

This link is yours. Please don't repost it — it is the whole shop.

Prefer to try first? The free Lite version is on GitHub, MIT licensed.

What is in it

10 subagents

bug-hunter finds the root cause instead of the symptom — it reproduces first, then traces the bad value back to where it first went wrong. code-reviewer reports real defects and skips style opinions. test-writer matches the framework you already use, runs the suite and pastes the real output. Plus refactorer, api-designer, sql-tuner, perf-profiler, dependency-auditor, migration-planner, docs-writer.

12 slash commands

/commit/pr/review/debug/fix-ci/tests/security/cleanup/explain/scaffold/onboard/changelog

They read your real diff before they act. /commit matches your repo's existing message style and refuses to commit a secret. /review greps for every caller of every changed signature — the check human reviewers always skip.

7 hooks — the part that can actually stop Claude

Hooks are the only piece of Claude Code that can block an action rather than politely ask.

Every hook in this pack was executed against real hook payloads before it shipped — that testing found and fixed two genuine bugs. The docs give you the exact commands to verify it yourself in two minutes:

printf '{"tool_input":{"command":"rm -rf /"}}' | ./guard-destructive.sh; echo $?
# 2 = blocked, 0 = allowed

4 skills

ship-it — the full pre-merge gate (build, typecheck, tests, review, security, leftovers, migration safety) that stops at the first blocker. incident-triage — production is down right now: stop the bleeding, then root-cause. spec-to-plan. legacy-rescue — writes characterisation tests before you touch untested legacy code, so you can tell a fix from a regression.

6 CLAUDE.md templates

Universal · Next.js + TypeScript · Python + FastAPI · Node API · Flutter · Monorepo. Each one with a commands table and a real Gotchas section.

A one-page cheatsheet

CLAUDE.md vs commands vs subagents vs skills vs hooks — which to use when, the exact frontmatter for each, every hook event, what exit code 2 actually does, and the precedence rules.

Install

unzip claude-code-drop-in-kit.zip
cd claude-code-drop-in-kit
bash install.sh            # this project
bash install.sh --global   # ~/.claude, every project

Then restart Claude Code. Requirements: Claude Code, bash, git, python3. Windows works under WSL.

Licence

One developer, unlimited personal and commercial projects, forever. Modify it freely. Don't resell or redistribute it. The free Lite subset on GitHub is MIT and you can do anything you like with that.


Built by Fayaz Bin Salam (@p32929). Questions or a bug? Open an issue.