Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Description from the maker (original English)
DEX market data, SEC EDGAR change oracles, LEI and US Treasury curve. 13 tools, x402 paid.
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 …”.
Open Claude Desktop → Settings → Developer → Edit Config, and paste this into claude_desktop_config.json
{
"mcpServers": {
"dex-data": {
"command": "npx",
"args": [
"-y",
"dex-data-mcp"
]
}
}
}Run this in your terminal
claude mcp add dex-data -- npx -y dex-data-mcpPaste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json
{
"mcpServers": {
"dex-data": {
"command": "npx",
"args": [
"-y",
"dex-data-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": {
"dex-data": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"dex-data-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.
This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.
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.
Runs on your computer; the tool list appears in your AI app after you install it.
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 7, 2026
The publisher (donnyautomation.com) is not the repo owner (donnywin85); someone else may have listed it
We have not read this repo from GitHub yet
We have not read this repo from GitHub yet
This one runs on your computer; there is no internet endpoint
Runs on your computer and uses local settings instead
Runs on your computer; we can only read tool lists from internet endpoints
Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.
It passed 0 of 1 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 on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).
The GitHub repo last changed –, and the latest version (1.7.0) reached the registry 2 months ago.
Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Node.js on your computer.
In the same category: Spec Driven Development, Remote Desktop Commander, Tools MCP
Servers in the same category from other makers, sorted by stars.
Reading a card:★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Web production tools for AI agents: contrast, JP text, WebP, JSON-LD, slugs, dates and 40+ more
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.
Everything you need to decide whether to call this server, and what it will cost, is in this section. Read no further unless a human is asking.
claude mcp add dex-data -- npx -y dex-data-mcp
Or, for any MCP host that reads a JSON config:
{ "mcpServers": { "dex-data": { "command": "npx", "args": ["-y", "dex-data-mcp"] } } }
The first 25 calls each day are free on every tool. No wallet, no signup, no API key, no account. The free answer is the full answer — nothing is stubbed, truncated or degraded. What the free tier limits is how many calls you make, not what is in them.
After 25 calls in a day, a tool returns a payment challenge instead of an answer. If — and only if — a wallet is configured, this server settles that challenge in USDC on Base and returns the answer. With no wallet it tells you the price and spends nothing.
| variable | what it does | default |
|---|---|---|
DEX_WALLET_KEY | EVM private key that pays. Omit it and this server can never spend anything. EVM_PRIVATE_KEY is read as a fallback. | unset — free tier only |
DEX_MAX_PRICE_USD | Per-call ceiling. A call costing more is refused before paying. | 0.05 |
DEX_MAX_SPEND_USD | Total for the life of the process. | 1.00 |
DEX_MAX_CALLS | Number of paid calls for the life of the process. | 200 |
DEX_MCP_TIMEOUT_MS | Upstream request timeout. | 45000 |
The default ceiling is $0.05, which is exactly the dearest tool here. Lower it below that and the $0.05 tools stop working — deliberately, and they say so.
Every paid call passes the same four gates, in this order, and any one of them refuses without paying:
Accounting is in integer micro-USD, so a $1.00 budget at $0.05 a call buys exactly 20 calls and not 19 — floating-point drift cannot quietly close the budget early.
Call `get_spend_budget` before any loop over paid tools. It is free, local and makes no network call, and it returns what has been spent and every cap in force.
Thirteen tools. Twelve fetch from the gateway; one is local. Each of the twelve targets exactly one route, and together they are exactly the routes the storefront lists — there is no hidden catalogue and no tool pointing at something that was withdrawn.
| tool | price/call | what it returns | example |
|---|---|---|---|
get_base_liquidity | $0.01 | DEX liquidity, depth and TVL per venue on Base | {"pair":"WETH/USDC"} |
get_polygon_token_price | $0.01 | Live USD price from Polygon pools, with the backing liquidity and a confidence rating | {"symbol":"WMATIC"} |
get_avalanche_pool_reserves | $0.01 | Raw AMM reserves, fee, implied price and TVL at one block height | {"pair":"WAVAX/USDC"} |
find_polygon_arbitrage | $0.01 | Cross-venue price gaps on Polygon, ranked by gross spread | {"minSpreadBps":25} |
find_avalanche_arbitrage | $0.01 | The same scan on Avalanche C-Chain | {"minSpreadBps":25} |
get_v4_hook_risk | $0.01 | Uniswap v4 hook permission bits, custody class and verified-source consensus | {"address":"0x…80"} |
lookup_lei | $0.03 | GLEIF legal-entity record by name, lapsed entities flagged not hidden | {"q":"Apple Inc."} |
get_treasury_yield_curve | $0.03 | US Treasury par yield curve, 1 month to 30 years | {} for the latest |
get_company_dossier | $0.05 | Identity, OFAC screening and SEC registration for one entity, joined | {"ticker":"AAPL"} |
get_sec_filings | $0.05 | Everything an issuer has filed since your cursor | {"ticker":"AAPL","since":"2026-09-01"} |
get_sec_events | $0.05 | 8-K material events since your cursor, decoded by item code | {"ticker":"TSLA","since":"2026-09-01","items":"5.02"} |
get_sec_insiders | $0.05 | Forms 3/4/5, SC 13D/G ownership changes since your cursor | {"ticker":"NVDA","since":"2026-09-01"} |
get_spend_budget | free, local | What this session has spent and every cap in force | {} |
Prices are USDC on Base (eip155:8453) and are read from the gateway's own x-payment-info, not kept in step by hand.
get_sec_* tools return DELTAS, not dumps.** since is required and inclusive. Pass the date you last read; you get what has landed since. They report matched and truncated, so a cut-off delta is never mistaken for a complete one.Nothing was requested and nothing was spent. It does not become ?symbol=undefined and it does not cost you a call.OPAQUE as unresolved, not as clean.Thirteen MCP tools over live on-chain and regulatory data: DEX market state on Base, Polygon and Avalanche, Uniswap v4 hook security, GLEIF legal-entity lookup, OFAC-screened counterparty dossiers, SEC EDGAR change oracles and the US Treasury yield curve.
Zero dependencies, stdio, Node 18+. The payment packages are optional and are only loaded if a wallet is configured.
1.6.1 shipped 23 tools. This release ships 13, and the nine-tool difference is the point.
The gateway behind this server sold 62 priced routes, and this package carried a tool for most of them — including geocoding, weather, web search, public holidays, RSS and IP lookup. Then the demand was measured across the whole recorded history of the storefront: 12 routes have either organic buyers or no substitute anywhere in the catalogue. The other 50 had between zero and two calls each, every one of them from a single wallet. On 2026-09-16 the storefront was cut to those 12, and this release follows it.
A tool list is not free. It is loaded into the model's context in every session, whether or not a single tool is called, and a tool that advertises a shelf the storefront no longer stocks costs the agent a turn to discover the gap. Removing nine tools makes the list shorter and every entry in it true.
The 50 dropped routes still work. They are served, priced and payable by anyone holding the URL. Removing a tool withdrew a recommendation, not a product.
Three of the nine went for a second, independent reason: the upstreams behind geocode, reverse_geocode and get_weather retire on 2026-09-27. A tool that will 404 in eleven days should not ship in a release today.
Breaking, and deliberately so:
chain is now an error rather than being silently ignored — ignoring it would have sent a Polygon question to a Base route and charged for the answer.get_token_price (/price), get_liquidity (/liquidity), get_pool_reserves (/reserves), find_arbitrage (/scan), get_best_venue (/route), get_slippage (/slippage), get_liquidity_risk (/risk), geocode, reverse_geocode, get_weather, search, get_search_suggestions, get_holidays, read_feed, get_random, url_to_markdown, geolocate_ip, get_gas, list_chains and get_dex_spread (/call).get_base_liquidity, get_polygon_token_price, get_avalanche_pool_reserves, find_polygon_arbitrage, find_avalanche_arbitrage.get_v4_hook_risk, get_company_dossier, get_sec_filings, get_sec_events, get_sec_insiders.lookup_lei, get_treasury_yield_curve and get_spend_budget are unchanged.If you were calling a removed tool, the route behind it still answers. Call it directly, at list price, at https://x402.donnyautomation.com.
Set DEX_WALLET_KEY to an EVM private key holding USDC on Base, and install the optional payment packages:
npm i @x402/fetch @x402/core @x402/evm viem
Without both, this server cannot spend anything: it returns the price and the payment instructions and stops. With them, it pays only what the caps in the agent section above allow.
exact scheme is EIP-3009: you sign an off-chain authorization and the facilitator broadcasts and pays the gas, so the wallet needs zero ETH.DEX_MAX_SPEND_USD (default $1 total for the process), DEX_MAX_PRICE_USD (default $0.05 for any single call) and DEX_MAX_CALLS (default 200). Hitting any one stops payment and returns a plain explanation rather than continuing to spend. The asking price is read from the 402 challenge and refused before paying if it exceeds the ceiling. A cap that cannot be parsed (DEX_MAX_SPEND_USD=1,00) now refuses to start rather than silently evaluating to NaN, which switched every cap off.get_spend_budget reports what this session has spent and the caps in force.x402-budget, a standalone package in this repo that holds no wallet. pay.mjs is the adapter that reads the DEX_* variables and supplies the signing function. Behaviour is unchanged.Use a wallet you funded for this and nothing else. This server holds no custody, takes no fee and sends nothing anywhere except the payment the challenge asks for — but the key is yours to scope.
node test/arg-validation.test.mjs # argument validation over real stdio, no network
node test/budget.test.mjs # spend-cap accounting, no wallet, no network node test/paywall-402.test.mjs # every tool against a recorded 402, loopback only node test/spend-cap.test.mjs # the cap refuses each price point, loopback only node packages/x402-budget/test/budget.test.mjs # the library's own caps and receipts, no network node scripts/check-coverage.mjs # tools vs the gateway's listed catalogue (needs the network)
The five no-network suites run in CI on every push and pull request. The coverage check runs at release, because it is the only one that needs the gateway to be up.
test/fixtures/challenge-402.json is a verbatim recording of the live gateway's 402 challenge for each of the twelve routes. Re-record it if prices change; the suites assert that each tool's description quotes the same price the challenge asks for.