# Stack Audit — install instructions (sample)

This is a sample install.md from a real Stack Audit deliverable. Same format
ships with every audit. ~60 seconds to drop in.

## Step 1 — confirm Claude Code is installed

```bash
which claude
# → should print something like /usr/local/bin/claude
```

If missing: `curl -fsSL https://claude.ai/install.sh | sh` or follow
https://docs.claude.com/en/docs/claude-code/quickstart

## Step 2 — drop the skills into ~/.claude/skills/

```bash
mkdir -p ~/.claude/skills/stripe-webhook-guard
cp skills/stripe-webhook-guard/SKILL.md ~/.claude/skills/stripe-webhook-guard/

mkdir -p ~/.claude/skills/rls-coverage-check
cp skills/rls-coverage-check/SKILL.md ~/.claude/skills/rls-coverage-check/

mkdir -p ~/.claude/skills/release-notes
cp skills/release-notes/SKILL.md ~/.claude/skills/release-notes/
```

(Audit deliverables ship with the actual files — this is the install pattern.)

## Step 3 — verify

In any Claude Code session inside your repo:

```bash
claude --print "list available skills"
```

You should see your three new skills in the output. Try one of the trigger
phrases (e.g. "deploying webhook") to confirm activation.

## Step 4 — optional hook script

If your audit included a `hooks/env-coverage.sh` or similar:

```bash
mkdir -p .claude/hooks
cp hooks/env-coverage.sh .claude/hooks/
chmod +x .claude/hooks/env-coverage.sh
```

Hooks run on commit/push — see `.claude/hooks/env-coverage.sh` comments for
the trigger event each one binds to.

## What "tested locally before delivery" means

Each skill in your audit deliverable was test-run on a clean clone of your
repo before shipping. The trigger phrases work, the file:line references are
current, and the suggested patches apply cleanly. If anything is off, reply
to the delivery email — fixes within 24h or your $49 back.

## Need help?

Email: agentstack.team@proton.me
Race log: agentstackhq.net/race
Free skill template: agentstackhq.net/free