Finance Toolkit
500+ transparent financial methods calculated from raw statements, not third-party endpoints.
Description from the maker (original English)
Trading-card research and seller economics: identify cards, check sold comps, and compare fees.
These three are never merged into one score: opinions, measurements and popularity are different things.
We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.
Pick the AI app you use, then copy the settings.
You need
Which app do you use?
Pick your app above. We remember it, so next time the button says “Install in …”.
Works in Claude on the web, desktop and mobile:
https://mcp.pulltrader.app/mcpRun this in your terminal
claude mcp add --transport http seller-economics https://mcp.pulltrader.app/mcpPaste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json
{
"mcpServers": {
"seller-economics": {
"url": "https://mcp.pulltrader.app/mcp"
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
Paste into .vscode/mcp.json in your project
{
"servers": {
"seller-economics": {
"type": "http",
"url": "https://mcp.pulltrader.app/mcp"
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
ChatGPT connects only to servers on the internet:
https://mcp.pulltrader.app/mcpMenu names can differ slightly by plan and over time.
Your choice stays on this device in a cookie; no account needed.
These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.
The server asks you to sign in before it lists its tools.
We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 9, 2026
The registry name was verified against the domain pulltrader.app (the registry checks domain ownership)
We have not read this repo from GitHub yet
We have not read this repo from GitHub yet
Responded in 1099 ms (HTTP 401)
Declares OAuth sign-in, so your AI app can take you to log in
Not checked yet; we read the tool list when we check the endpoint
Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.
It passed 3 of 3 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.
The registry does not list a required key, though some servers ask you to sign in on first use.
It runs in the maker’s cloud. Connect over the internet with nothing to install, but what you send passes through the maker’s servers.
The GitHub repo last changed –, and the latest version (0.6.1) reached the registry 36 days ago.
Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. It connects by URL, so apps that accept remote MCP servers work too, such as ChatGPT in developer mode.
In the same category: Finance Toolkit, SolidInvoice, Korea Tax Calculator
Servers in the same category from other makers, sorted by stars.
Reading a card:↓ Downloads per weeki★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi
500+ transparent financial methods calculated from raw statements, not third-party endpoints.
Create and send invoices and quotes, manage clients, and track payments in SolidInvoice.
Korean tax calculation and tax-audit judgment MCP: 96 tools with statutory citations.
Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.
Identify a card, pull recent sold comps, summarize its market value, chart its price history, and compare what a seller keeps across eBay, Pulltrader, and other card marketplaces — inside your AI assistant.
A public, read-only Model Context Protocol server for trading-card research and selling. Eight tools: a Kapture stolen-card registry check, four back onto Pulltrader's card data (identification, comparable sold sales, market value, price history), three run Pulltrader's deterministic seller-economics engine over a dated, versioned fee schedule.
Market figures are estimates derived from recent sales. Fee math is deterministic — the server computes it and never asks the model to infer fees from prose. eBay figures are clearly labeled estimates.
POST /mcp)https://mcp.pulltrader.app/mcpapp.pulltrader/seller-economics2025-06-18 (the client's requested version is echoed when present)Trading-card sellers, dealers, shops, and collectors who want to know what a card is, what it's worth, and what they'd actually take home selling it — without leaving their assistant.
| Does | Doesn't |
|---|---|
| Resolve a prose card description into structured fields | Identify a card from an image (text only) |
| Return recent comparable sold sales (price + date) | Return listing, affiliate, or per-sale outbound URLs |
| Summarize market value: median, mean, p10–p90, volatility, confidence | Claim a guaranteed value or give financial advice |
| Chart price history by day / week / month with a trend | Make any write, order, or account change |
| Estimate seller fees and net proceeds per selling method | Claim one platform is universally cheapest |
| Solve for the price needed to hit a target net | Model auction formats (hammer, buyer's premium, consignment) |
| State its assumptions, data freshness, and limitations | Cover non-card categories or currencies other than USD |
check_kapture_cert accepts { "cert": "00123456" } and preserves leading zeros. CLEAN only means no verified stolen report matched. STOLEN includes the Kapture report and a voluntary Pulltrader recovery-form link. The tool never collects personal information, infers possession, or submits a recovery report. The backend holds KAPTURE_API_KEY; no new Worker secret is needed. Recovery handoff is manual until Kapture provides a reporting API. See pulltrader-backend/docs/KAPTURE.md.
These bridge to the Pulltrader backend. Card data is limited-public: a capped sample of sales, no per-listing or affiliate URLs, and only the catalog reference image.
Each accepts either query (a natural-language description, e.g. "2023 Panini Prizm Victor Wembanyama #136 Silver PSA 10") or item (structured fields: player_athlete, year_manufactured, set_name, card_number, parallel_variety, grader, grade, sport). One of the two is required.
| Tool | What it returns | ||
|---|---|---|---|
identify_card | Canonical fields (player/athlete, year, set, number, parallel, grader, grade, category) plus a confidence level and which fields resolved. Takes query only. Does not price the card. | ||
search_card_sales | A capped sample of recent comparable sold sales (price + date) plus a market snapshot. Optional limit. | ||
summarize_card_market | Canonical market value: median, mean, p10–p90, volatility, sample size, confidence. | ||
get_card_price_history | A chart-ready time series with a trend. Optional interval: day \ | week \ | month. |
When eBay has no dated comps, search_card_sales and summarize_card_market fall back to the authoritative vendor price and surface market_value, value_source (TCG Market \| CardSightAI \| eBay Comps), price_change_7d, price_change_30d, and a single catalog image.
| Tool | What it returns |
|---|---|
compare_selling_costs | Estimated fees and net proceeds for one sale across selling methods, with a per-method fee breakdown, the difference vs the eBay baseline, and the assumptions used. Requires sale_price. |
calculate_required_sale_price | The per-item price needed to reach a target take-home (or net profit, when acquisition_cost is given) on a single method. Requires target_net. |
explain_selling_method | Plain, structured explanation of how each method owns the listing, fulfills, and charges fees. Derived from the same engine, so it never drifts from compare_selling_costs. |
Shared inputs (compare_selling_costs, calculate_required_sale_price):
| Field | Type | Default | Notes | |||
|---|---|---|---|---|---|---|
currency | enum USD | USD | Only USD supported | |||
quantity | integer | 1 | Per-order fixed fees applied once | |||
shipping_amount | number | 0 | Affects eBay's fee base | |||
item_category | enum trading_cards | trading_cards | Only trading cards | |||
seller_plan | free\ | starter\ | pro\ | shop | free | Marketplace payout tier |
seller_level | enum | derived from plan | Explicit payout level override | |||
seller_covers_fees | boolean | false | If false, the buyer pays the platform fee | |||
ebay_store_subscription | boolean | false | eBay Store rate (12.35%) vs individual (13.25%) | |||
acquisition_cost | number | — | Switches output to net profit | |||
ebay_fee_percent_override | number | — | Override the estimated eBay FVF % (flat) |
Supported methods: ebay, pulltrader_marketplace, pulltrader_fbp, pulltrader_storefront, pulltrader_pos, plus estimated competitor marketplaces tcgplayer, manapool, misprint, fanatics_collect, goldin (off by default — opt in via methods). Competitor figures model fixed-price / Buy Now seller fees only.
On a $250.00 trading-card sale, estimated seller proceeds by method:
For these inputs, Your Pulltrader storefront (you ship) returns about $33.53 more than Sell it yourself on eBay. Estimate based on fee schedules updated 2026-06-01 (Pulltrader) and 2026-06-27 (eBay, estimated). Estimates only, for trading cards in USD. ...
eBay = 13.25% individual FVF + $0.40 per-order fee (orders over $10). The storefront keeps 100% because Pulltrader has no seller fee on storefront/POS — only the platform fee, paid by the buyer here. The platform fee is always charged on Pulltrader card sales (buyer pays by default).
Settings → Connectors → Add custom connector → URL https://mcp.pulltrader.app/mcp.
Settings → Apps & Connectors → Advanced → enable Developer mode → Create → URL https://mcp.pulltrader.app/mcp.
Add to your MCP client config:
{
"mcpServers": { "scout-by-pulltrader": { "url": "https://mcp.pulltrader.app/mcp" } } }
Tested clients: Claude (web/Desktop), ChatGPT developer mode, Cursor, MCP Inspector. We do not claim universal compatibility with every assistant.
The /mcp endpoint is public and read-only. A separate seller resource is implemented at /seller/mcp; OAuth is provisioned and its signed-in pilot is pending. See Seller MCP.
| Limit | Default | Notes |
|---|---|---|
All POST /mcp | 60 / IP / min | HTTP 429 + Retry-After. Fail-open if KV is down. |
| Data tools (identify / comps / market / history) | 8 / IP / min, 40 / IP / day | Tool isError RATE_LIMITED. Fail-closed if KV is down. |
| Data tools global | 10,000 / day | Circuit breaker for model + comps spend. |
| Batch size | 5 messages | Larger batches are rejected. |
Requires the MCP_ABUSE KV namespace (see wrangler.toml). Tune via PUBLIC_RATE_LIMIT_PER_MIN, DATA_RATE_LIMIT_PER_MIN, DATA_RATE_LIMIT_PER_DAY, DATA_GLOBAL_LIMIT_PER_DAY.
Input problems come back as a tool result with isError: true and a stable code, not a JSON-RPC error: INVALID_INPUT, NOT_FOUND, RATE_LIMITED, UPSTREAM_ERROR, UPSTREAM_TIMEOUT, DATA_BACKEND_UNAVAILABLE, INTERNAL_ERROR. Unknown methods and malformed envelopes use standard JSON-RPC error codes.
The four card tools require the backend bridge (PULLTRADER_API_BASE + SCOUT_MCP_SECRET). Without it they degrade to DATA_BACKEND_UNAVAILABLE; the three seller-economics tools are pure and always available.
Current schedule versions are served live at /version.
npm install
npm run typecheck # tsc --noEmit npm test # vitest npm run dev # wrangler dev (local)
Deploy and registry submission require explicit approval — see docs/DEPLOYMENT.md and docs/DISTRIBUTION.md. Do not deploy or publish from a development session.
wrangler kv namespace create MCP_ABUSE
wrangler kv namespace create MCP_ABUSE --preview
wrangler secret put SCOUT_MCP_SECRET
/seller/mcp; global per-user quotas remain future work.Built and maintained by Pulltrader. Support: support@pulltrader.app
This repository is the public mirror of the Pulltrader MCP Worker, exported from the Pulltrader monorepo on each release. Issues and questions are welcome here; pull requests are applied upstream. wrangler.toml KV ids are redacted — deploys run from the monorepo.