gold-ticker-live-audit

name	gold-ticker-live-audit
description	Use for any broad audit, planning, PR-prep, or repo-understanding task. Generates a coherent picture of the product across pricing truth, frontend/mobile, SEO, automation, security, and release readiness.
when_to_use	["Onboarding to the repo for the first time in a session","Planning a multi-surface change (e.g. revamp, redesign, expansion)","Pre-PR sanity sweep before opening a draft","Audit-only PRs that produce a report instead of code"]
related_files	["AGENTS.md",".github/copilot-instructions.md",".github/instructions/*.instructions.md","docs/AI_AGENT_OPERATING_SYSTEM.md","docs/REVAMP_PLAN.md"]
related_prompts	[".github/prompts/pr-review.prompt.md",".github/prompts/release-readiness.prompt.md"]

Skill: Gold Ticker Live Audit

The audit skill is the entry point for understanding the product, planning a change, or reviewing a PR. It produces a coherent picture across all the surfaces that matter at Gold Ticker Live.

When to use

• Starting a session with a vague or large prompt ("revamp UI", "improve SEO", "tracker overhaul")
• Auditing before opening a draft PR with a plan file
• Pre-merge sanity sweep on a non-trivial PR
• Triaging a regression spotted in production

Project snapshot to load first

What	Where
Product identity	AGENTS.md project overview, .cursor/rules/*.mdc
Pricing rules	.github/instructions/gold-pricing.instructions.md
Visual identity / tokens	styles/global.css, docs/DESIGN_TOKENS.md
Tracker state	docs/tracker-state.md
Active revamp tracks	docs/REVAMP_PLAN.md
Recent commits / known issues	git log --oneline -30, docs/REPO_AUDIT.md
Workflow health	.github/workflows/, docs/X_AUTOMATION_OBSERVABILITY.md

Required repo inspection (in this order)

1. AGENTS.md (full read)
2. .github/copilot-instructions.md (full read)
3. The relevant .github/instructions/*.instructions.md for the surface(s) you're touching
4. The page(s) you'll edit: HTML + matching styles/pages/*.css + matching src/pages/*.js
5. docs/REVAMP_PLAN.md — is your change already on the plan?
6. package.json scripts — know what validate / quality / test actually run

Workflow

1. Restate the goal in one sentence. If it's vague, expand it to a measurable spec.
2. Identify surfaces touched. Map to .github/instructions/*.instructions.md.
3. Run the relevant checklists below.
4. If the task is >2h of work, write a plan file under docs/plans/YYYY-MM-DD_<slug>.md and open a draft PR with just the plan.
5. Implement in coherent commits, one surface at a time.
6. Verify with AGENTS.md core commands + testing-qa.instructions.md.
7. Report using .github/prompts/_output-format.md (severity, file, impact, exact fix).

Checklists in this skill

• checklists/preflight.md — before touching any code
• checklists/pricing-truth.md — formula / freshness / labels
• checklists/frontend-mobile.md — mobile + RTL + tokens
• checklists/seo-governance.md — canonical + sitemap + noindex
• checklists/automation-safety.md — workflows + X posting
• checklists/release-readiness.md — merge gate

Common mistakes (from prior sessions)

• Reading the README and skipping AGENTS.md.
• Treating the repo like a generic static site (ignoring pricing truth + freshness labels).
• Shipping a "tracker tweak" that strips the freshness pill.
• Editing sitemap.xml directly (next build wipes it).
• Hand-coding karat factors outside src/config/karats.js.
• Forgetting to delete playwright-report/ before npm test / npm run validate.
• Treating post_gold.yml as a draft workflow when it's live in production.
• Claiming "tests pass" without naming the command and the count.

Final report template (audit-only PR)

# Audit Report — <scope>

## Summary
<2–4 sentence executive view>

## Surfaces inspected
- <surface>: <file paths> → <state>

## Findings
### Blocking
- <issue>: <evidence> → <recommended action>

### Important non-blocking
- ...

### Nice-to-have
- ...

## Recommended next PR(s)
- <title> — <scope> — <expected effort>

## Verification run
- `npm run lint` → PASS/FAIL
- `npm test` → PASS/FAIL, N tests
- `npm run validate` → PASS/FAIL
- Manual: <viewports, RTL spot, screenshots>

## Risks
- <known unknowns>