no-getbets - SKILL.md Agent Skill

name: no-getbets description: "Hard ban on any reference to the external domain getbets.us on cruisinginthewake.com. Enforced via PreToolUse hook (.claude/hooks/no-getbets-guard.js) and git pre-commit hook (.githooks/pre-commit). Fires automatically on every Write/Edit/MultiEdit and every commit; no manual invocation." version: 1.0.0

no-getbets — Hard Ban on External Domain Reference

The rule

The string getbets (case-insensitive) must never appear in any production-facing file on cruisinginthewake.com. This includes the published domain, any path served to users, and any source file that compiles to a served path.

No exceptions for disclaimers, "not affiliated with" notices, comparison tables, or any other framing. The ban is absolute on production surfaces.

Why

getbets.us is an external US online-casino review/affiliate site, flagged as suspect by Scam Detector (reference). It has no legitimate connection to cruise content and no public web page genuinely pairs the two domains.

On 2026-05-28, AI grounding analytics surfaced 226 LLM citations on the queries casino royale royal caribbean getbets.us (196) and casino royale royal caribbean getbets (30). The site's own restaurants/casino.html was clean — zero references. The pattern was identified as one of:

Black-hat SEO probe: getbets.us appending its domain to high-volume cruise queries to siphon ranking authority through co-mention
Bot/scraper-generated query template propagated through content-farm infrastructure
Affiliate spam network embedding the pairing in low-quality anchor/footer text

The diagnosis: adversarial co-mention, not organic demand. Any defensive reference to getbets.us added to this site — including a disclaimer — would strengthen the SEO pairing the adversary is trying to create. The right defense is structural absence.

Exempt paths

Documentation about the ban itself, and the private admin tooling that defends against it, are the only legitimate contexts for the string getbets in this repo. The following paths are exempt:

.claude/skills/no-getbets/** (this skill)
.claude/hooks/no-getbets-* (the enforcement hook)
.claude/standards/no-getbets* (any future standard entry)
.githooks/pre-commit (the git enforcement)
CLAUDE.md, SKILLS.md, AGENT.md (rule discoverability)
audit-reports/** (investigation artifacts)
admin/check-no-getbets* (any admin tooling for the check)
admin/seo/** (Search Console disavow files and related defensive SEO artifacts — never served)

The exempt list is duplicated in .claude/hooks/no-getbets-guard.js (EXEMPT_PATH_PATTERNS) and .githooks/pre-commit (path-filter grep). Any new exemption must be added in both places.

Enforcement (defense in depth)

Layer	Where	When it fires	Behavior on hit
Claude Code PreToolUse hook	`.claude/hooks/no-getbets-guard.js`	Every Write / Edit / MultiEdit	Denies the tool call with a denial reason
Git pre-commit hook	`.githooks/pre-commit`	Every `git commit`	Blocks the commit with the list of offending files
This skill	`.claude/skills/no-getbets/SKILL.md`	Discovered at session start	Documents the rule and rationale for human + agent readers
Cognitive memory	`inthewake` scope, protected	Semantic recall in any session	Returns the rule when sessions search for related terms
Top-level policy	`CLAUDE.md` "Hard bans" section	Every session reads at start	Declarative rule before any work begins

Operating instructions

Never add the string getbets to any production file. No disclaimer, no comparison, no defensive notice.
If a hook blocks a legitimate edit (e.g., extending this skill itself), add the file's path pattern to EXEMPT_PATH_PATTERNS in .claude/hooks/no-getbets-guard.js AND to the path filter in .githooks/pre-commit. Both must agree.
If the adversarial campaign escalates (citation count grows substantially, real referral traffic begins appearing from getbets.us), the next layer is a Search Console Disavow filing — outside the scope of this skill, but the trigger for that escalation is documented here.
Do not bypass the git hook with --no-verify for any commit touching production files. The bypass is reserved for cases where the documentation paths themselves are being edited.

Verification

To confirm the ban is intact across the codebase:

grep -rli 'getbets' \
  --exclude-dir=.git \
  --exclude-dir=node_modules \
  --exclude-dir=audit-reports \
  --exclude-dir=.claude/skills/no-getbets \
  --exclude=.claude/hooks/no-getbets-guard.js \
  --exclude=.githooks/pre-commit \
  --exclude=CLAUDE.md \
  --exclude=SKILLS.md \
  .

A clean run produces no output. Any hit is either a violation or a legitimate documentation file that needs adding to the exempt list.

Companion files

.claude/hooks/no-getbets-guard.js — Claude Code PreToolUse enforcement
.githooks/pre-commit — git commit enforcement (check #5)
CLAUDE.md — top-level policy declaration
SKILLS.md — skill index registration
Cognitive memory inthewake scope — protected entry

Defense is structural absence, not defensive presence. The string never appears, so no SEO signal can form.