card-profile-recommend（卡资料推荐）

Card Profile Recommend

Return a graded portfolio audit plus opinionated new-card recommendations with signup bonus strategy, churning paths, and issuer rule checks.

When To Use

When the user wants to know which cards to keep, drop, or add next. Trigger phrases: "card-profile-recommend", "recommend cards", "what cards should I add", "credit card recommendations", "optimize my wallet", "what card should I get next", "improve my card lineup".

Input Format

The user provides a comma-separated list of cards they currently hold:

• card-profile-recommend Chase Sapphire Preferred, Amex Gold, Citi Double Cash

Optional: opening dates per card (inline or separate list):

• card-profile-recommend CSP (Jan 2024), Amex Gold (Mar 2023), Double Cash (2021)

When opening dates are provided, calculate exact 5/24 count and factor into grading decisions.

Workflow

1. Parse card list from comma-separated input.
2. Resolve each card — normalize and match to exact variants. If any card is ambiguous, return a numbered choice list for that card and stop.
3. Search — use WebSearch by default for each card plus any needed gap-category searches. If BRAVE_API_KEY is available and curl exists, you may use Brave Search API instead, but do not burst requests blindly.
4. Fetch pages — for each card, fetch the top issuer URL + 1 secondary (prefer thepointsguy.com). For new-card candidates, fetch up to 2 secondary pages.
5. Pace any follow-up searches — if more searches are needed, serialize them with short waits rather than firing them all at once.
6. Collect — for each card: annual fee, statement credits (with conditions), earning categories with rates, welcome offer status, notable benefits.
7. Portfolio economics — compute total gross fees, total claimable credits, net annual cost. Per-card net cost.
8. Grade each card — MVP / Keep / Consider Dropping per grading criteria below.
9. Point valuations — determine effective cpp for each currency using TPG valuations + transfer-access rule.
10. Earning map — best card per category with effective value (rate × cpp).
11. Identify gaps — categories earning below 2x / 2%.
12. Recommend 2–3 new personal cards — fill gaps, add ecosystems, maximize SUB windows, include churning strategy.
13. Build application sequence — ordered by priority with timing notes.
14. Confidence — flag uncertain claims.

Step 1: Card Identity Resolution

Common Abbreviations

Supported Issuers

American Express, Bank of America, Barclays, Bilt, Capital One, Chase, Citi, Discover, Robinhood, U.S. Bank, Wells Fargo.

Step 2: Search

Use the platform's WebSearch and WebFetch tools by default. If BRAVE_API_KEY is available and the runtime also provides curl, you may use Brave Search API instead for faster and more repeatable search results.

Optional Brave template:

curl -sS "https://api.search.brave.com/res/v1/web/search?q=CARD+NAME+benefits+credits+annual+fee&count=10" \
  -H "X-Subscription-Token: $BRAVE_API_KEY"

Do not assume any search provider tolerates a large burst of parallel searches.

Search Budget Rule

Treat search as scarce and paced. Built-in web search is the default path; if Brave mode is used, it may rate-limit after only a few closely spaced requests.

• Start with the most important cards first.
• Fetch issuer and approved secondary pages before deciding whether more searches are needed.
• When multiple searches are required, serialize them in small batches or add short waits of about 2 to 5 seconds between bursts.
• If Brave returns 429, wait about 8 to 15 seconds and retry once for the still-missing search.
• If Brave is unavailable, continue with WebSearch + WebFetch.
• If it still fails, continue with the best evidence already gathered and note the limitation in ## 🔍 Confidence Notes.

Additionally, search for new-card candidates targeting gap categories.

Fetch Pages

For each card, fetch the top issuer URL + 1 secondary with WebFetch.

An approved secondary page means a URL whose hostname matches an approved secondary domain used by this skill, such as thepointsguy.com for card cross-checks. Do not fetch or cite secondary pages from any other domain.

URL Safety Rules

• Prefer WebFetch for page retrieval. Use curl only for the optional Brave Search API calls above, not for arbitrary result URLs.
• Never execute a shell command that interpolates a raw URL taken directly from search results.
• Only fetch URLs when all of the following are true:

1. scheme is https 2. hostname matches a supported issuer domain or an approved secondary domain from this skill 3. the URL is being passed to WebFetch, not inserted into a shell pipeline

• If a result URL fails those checks, skip it and use the next valid result.

Issuer Domains (for classifying results)

Grading Criteria

MVP

• Net cost ≤ $50, OR unique 3x+ earn category, OR unique high-value benefit (lounge, hotel status, primary travel insurance)
• Best rate in at least one major spend category
• Not rendered redundant by another card in wallet

Keep

• Partially justifying benefit or earn rate
• Long credit history
• Transferable points diversification
• Under 12 months old

Consider Dropping

• Fees exceed claimable credits with no unique benefit
• All categories duplicated at equal or better rate
• No-fee downgrade path exists
• Never grade Consider Dropping if the card is the sole source of a transferable-points program

Unused Card Check

After building the earning map, flag any card that does not appear as "Best Card" in any category and has an annual fee.

Point Valuation: Transfer-Access Rule

A currency is only worth full TPG value if the wallet has a card that unlocks transfers. Without one, value is 1.0¢ (cash back).

Recommendation Logic

Select 2–3 new personal cards only. Never recommend business cards. Priority order:

1. Fill highest-value spend gap
2. Add new transferable-points ecosystem
3. Maximize signup bonus window (Chase-first if 5/24 allows)
4. Avoid worsening overlap
5. Prioritize elevated offers
6. Diversify issuers
7. Churning value — favor cards with SUB ≥ 5x annual fee, no-fee downgrade paths, and reasonable bonus cooldown windows. Note churn path explicitly.

Issuer Rules Reference

Required Output Sections

## 🃏 Cards Entered

Echo back every card with resolved full official name and opening date if provided.

## 📊 Portfolio Summary

Total cards, gross fees, claimable credits, net annual cost. Note credit utilization assumptions.

## 🏅 Card Grades

Each card graded MVP / Keep / Consider Dropping with one-line rationale. Always use full official card names. Sort MVP first, then Keep, then Consider Dropping.

## 🗺️ Earning Map

Table: Category, Best Card, Rate, Currency, CPP, Effective Value. Use full official card names. CPP reflects transfer-access rule.

## 🔻 Consider Dropping

Only when applicable. Per card: fee drag, what would be lost, downgrade path. Flag unused cards (not winning any earning map category). Omit entirely when all cards grade MVP or Keep.

## 🕳️ Portfolio Gaps

Numbered list of categories earning below 2x / 2%.

## ➕ Recommended Additions

2–3 new cards. Per card: name, welcome offer, annual fee, why it fits, net first-year value, issuer rule status, priority label, churn path if applicable.

## 🎯 Signup Bonus Strategy

Ordered application sequence with timing and spend feasibility.

## ⚖️ Issuer Rules Check

Only rules relevant to recommended cards. Bold blocking rules.

## 🔍 Confidence Notes

Flag uncertain, unconfirmed, or conflicting claims.

## 🔗 Sources

Numbered list of URLs fetched, as markdown hyperlinks with "Site - Topic" labels.

Output Rules

• Use one emoji per section heading.
• Always use full official card names (e.g., "Chase Sapphire Reserve" not "CSR").
• Use numbered lists for list-heavy sections.
• Keep content to condensed facts — no prose padding.
• Omit Card Identity section when all matches are confident.
• Do not show YAML blocks in output.

Confidence Definitions

• confirmed: supported by issuer terms or multiple approved sources
• unconfirmed: plausible but not fully resolved
• conflicting: sources disagree on a material fact