Indian credit card rules for AI agents: which card to use, caps, transfers, bank changes. All dated.
Scored across 5 tools
Each tool targets a clearly distinct operation: card detail, card discovery, rule-change history, spend recommendation, and transfer-route lookup. Although get_card includes transfer and earn data that also appears in transfer_partners and recommend, the descriptions make the boundaries clear.
All names use snake_case and are readable, but the verb patterns are slightly mixed: get_card, list_changes, search_cards, and transfer_partners are verb_noun/noun-like, while recommend is a bare verb. This is a minor deviation rather than a real inconsistency.
Five tools is well-scoped for an Indian credit-card intelligence server, with each tool covering a distinct workflow: search, detail, changes, recommendation, and transfers. There are no redundant or obviously unnecessary tools.
The surface covers the core read-only lifecycle: discovery, full card records, rule changes, spend recommendations, and transfer partners. Minor gaps remain around side-by-side card comparison and best-card-for-spend recommendations across all cards rather than only cards a person holds.
Everything on record for one card: fees and the waiver, earn rules with their caps, payment rail (UPI or card) and validity, exclusions, transfer partners with ratios and times, lounges, milestones, the statuses it grants, and its dated change history. Every row carries its grade, check date and source.
Inputs: slug (required).
Changes banks have made to card rules since a date: devaluations, fee changes, lounge gates, transfer partners added or removed, improvements. Newest first, upcoming ones included and marked. Each change carries the date it takes effect, the cards it touches, grade, check date, source and its own page on getbrief.cards.
Inputs: since (required), card, limit, until, +2 more.
Ranks the cards a person holds for one spend: a merchant (such as "MakeMyTrip", "Swiggy", "amazon.in") or a category, and an amount in rupees. Each card comes with the points, their ₹ value at Brief’s estimate and at the baseline its card page prints, the working, the rule and its source, any cap that bites, the transfer route the value assumes, and a cheaper channel (a bank’s travel portal) with the portal rule’s own condition, beside every portal rule weighed for that card (channels). When the order changes at baseline values, the answer says so. Month-to-date spend counts as zero unless wallet.month_to_date carries it, so a monthly cap reads as untouched; the answer says which. Nothing in the wallet is stored.
Inputs: wallet (required), amount_inr (required), upi, category, +2 more.
Finds Indian credit cards by name, issuer, network, fee band and three filters: forex markup of 1% or under, lounge access, transfer partners. Each card comes with its slug, annual fee, base earn, one line of facts, grade, check date, source and its page on getbrief.cards. Cards no longer issued are left out unless include_discontinued is true. Use a slug from here in every other call.
Inputs: limit, query, issuer, lounge, +5 more.
Where points can move, by programme or by card. A programme ("Marriott Bonvoy", "KrisFlyer", "Axis EDGE Miles", "HDFC Reward Points") returns the routes out of it and into it; a card slug returns the routes its points take. Each route carries the ratio as from:to, the transfer time, the minimum, any annual limit or card restriction, grade, check date, source and the transfer page on getbrief.cards.
Inputs: card, programme, include_closed.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Give your AI agent a spending limit: approval controls and single-use virtual cards.
India's financial MCP — cards, loans, FDs, insurance, MFs; EMI/SIP/tax calcs; NAV lookup; RAG.
Prepaid virtual cards for AI agents: one-time cards, spend caps, human approvals.
Payment infrastructure for AI agents: spending rules, approval flows, single-use virtual cards.
Add one secure layer between your agents and this server.
