We are now officially live on @ponsdotfamily.
Introducing Pons MCP : Autonomous payments for AI agents. MCP server + SDK for the Machine Payments Protocol. Settlement runs transparently on-chain, in USDG, on Robinhood Chain 4663.
Pons MCP bridges AI agents with Stripe's Machine Payments Protocol, enabling autonomous service payments with transparent Robinhood Chain settlement.
CA : 0x15da2596F4C21227185466066Bf0f19d9D526B8a
Web : https://t.co/0gJgD3NkrP
Github : https://t.co/0kzdKTGkQj
The PonsMCP TypeScript SDK and MCP server are designed to be inspectable and easy to integrate.
Developers can connect PonsMCP to their existing agent runtime and enable MPP payments with just a few lines of code. The server runs locally over stdio, making it compatible with MCP hosts such as Claude, Cursor, and other agent runtimes.
Spending controls are part of the payment flow itself. Per-transaction and daily limits are checked before funds move, and balance is checked before broadcast.
https://t.co/2DsQCMITU7
We have doing another $MCP Buybacks and burned... this round almost 1.5% supply is gone with 8.37% of the total supply burned permanently.
tx proof : https://t.co/SeZG3Qrdab
Live pons v2 launch intelligence.
Live on-chain snapshot of the pons v2 launch factory computed at unlock time: config count, latest block, factory address. Fresh on every request.
- Real chain read at unlock
- Factory config count
- Block-height proof
https://t.co/cuJtM87fMK
Testing PonsMCP Mission Control in beta.
The web console for agent payments on Robinhood Chain: configure, preview, and verify what your AI agent spends.
Not a wallet app, not a swap UI. 🧵
Top of the console, a live stat strip:
- $MCP price from the DEX, refreshed every 30s
- Latest block
- Network gas in gwei
- 25 MCP tools loaded
All read directly from chain. No mock data.
Payment flow:
1. Pick a service from the catalog, fields auto-fill
2. Policy guard checks pricing against the 100 USDG/tx cap
3. Quote returns the exact JSON your agent executes
Execution stays in your PonsMCP server. The browser never signs.
Services in the catalog:
- Sandbox payment: 0.01 USDG
- Grade A signals digest: 0.05 USDG
- Stock screener: 0.10 USDG
- Launch intelligence: 0.30 USDG
- SDK quickstart: 0.50 USDG
Each one unlocks computed data after on-chain receipt verification. Sandbox is the cheapest way to test your rail end to end.
Market Intelligence tab, fed by a live Grade A / Early Watch pipeline.
Per token: MC, ATH drawdown, liquidity, holders, rug score, smart money flow. Click any token for the full analysis.
Trade preview quotes from live pool reserves.
$MCP is live on Robinhood Chain (4663).
CA : 0x15da2596F4C21227185466066Bf0f19d9D526B8a
Name, symbol, and supply are verifiable on-chain. The console stat card shows live price and 24h change, linked to the pair.
https://t.co/qmYKlbxpqW
PonsMCP is an integration plugin that enables AI agents to execute autonomous payments using Stripe's Machine Payments Protocol (MPP) with transparent on-chain settlement via Robinhood Chain. Ship agents that discover, negotiate, and pay for services without human intervention.
What is PonsMCP?
PonsMCP bridges three critical technologies to enable autonomous agent commerce:
1. Stripe's Machine Payments Protocol: Industry-standard payment requests (HTTP 402) for machine-to-machine commerce
2. Robinhood Blockchain: Sub-second finality with sub-cent fees for transparent on-chain settlement
3. AI Agents: Autonomous execution framework for complex multi-step workflows Instead of forcing agents to navigate human-centric payment flows (checkout pages, billing forms, credit card entry), PonsMCP provides a standardized interface for discovering MPP-enabled services, authorizing stablecoin payments, and verifying delivery receipts.
https://t.co/2DsQCMITU7
Payment settlement is USDG, not PONS. PONS is the ecosystem asset of the launchpad; an agent that needs to pay for a service pays in a stable. Launch-tool research (should I buy this launch? what would the tax be? what does the creator have claimable?) is a decision layer — the execution of any launch trade stays in your own wallet, deliberately outside this SDK.
For launch creators: pons_escrow_balance reads the v2 fee escrow (balanceOf / balanceOfToken) so an agent can monitor claimable creator fees. Claiming itself (claim() / claimToken() on the escrow) is a wallet action — PonsMCP never moves someone else's fees.
Policy-safe MPP payment requests for autonomous agents. Quotes in USDG, execution from your PonsMCP server, proof from the chain.
https://t.co/iEOXhuVL3C
Your agent runtime does, via PONSMCP_PRIVATE_KEY. The browser console never receives a private key and never signs. Read-only tools work with no key at all.
A Model Context Protocol server and TypeScript SDK. Agents call tools to read pons launch data, quote payments, and settle in USDG with verifiable receipts. The server runs locally over stdio — plug it into Claude, Cursor, or any MCP host.
PonsMCP fully connected to Claude Desktop via MCP.
An AI agent that can screen tokenized stock markets, analyze memecoin launches, and execute on-chain payments in USDG — all through natural language.
No dashboard switching. No manual transactions.
Here's what happened 🧵
---
The setup:
@ ponsmcp SDK running as an MCP server inside Claude Desktop.
Just two credentials:
- Robinhood Chain RPC API key
- Wallet private key
Claude automatically discovers all 25 available tools on startup.
From there, everything happens through natural language prompts — from querying live chain data to executing transactions.
---
First test: tokenized stock screening.
Asked Claude to scan all 19 verified stock tokens on Robinhood Chain and rank them by liquidity.
Here's what it returned:
META: $4.35M liquidity, +0.69%
NVDA: $2.93M liquidity, +2.07%
TSLA: $860K liquidity, +3.95%
PLTR: $621K liquidity, +21.43%
SNDK: $266K liquidity, +24.28%
Live DEX data. Pulled directly from on-chain trading pairs.
All assets are tokenized debt securities, not equity shares.
---
Second test: Pons launch scanner.
Claude scanned 48 active launches and ranked them based on graduation progress and market activity.
One result stood out in this snapshot: SHROUD CARD.
→ 16% toward the 4.2 ETH graduation threshold
→ +105% 24h change
→ Active buys detected
Then Claude generated a research brief covering utility, deployer information, social links, and DEX liquidity.
One prompt. Multiple tools working together.
---
Final test: an actual on-chain payment.
The execution flow:
`pons_quote → pons_send_token → pons_tx_status`
Result:
✅ 0.01 USDG transferred
✅ Confirmed on-chain
✅ Single-block confirmation
Transaction:
`0xb2f79a554ecba2d2a3188ef4a902943201bad9a73a72deca9a2ad33ba2891ea2`
Gas used: 39,397
The entire workflow — quote, sign, broadcast, and verify — happened through Claude.
No wallet UI interaction required.
---
This is what agent-native infrastructure enables.
Not just an AI that explains what you could do.
An AI agent that can:
- Read live blockchain data
- Analyze token markets
- Discover and research launches
- Execute on-chain payments
- Verify transaction receipts
All through a single MCP integration.
ponsmcp × @AnthropicAI
From natural language to verified on-chain execution.
PonsMCP fully connected to Claude Desktop via MCP.
An AI agent that can screen tokenized stock markets, analyze memecoin launches, and execute on-chain payments in USDG — all through natural language.
No dashboard switching. No manual transactions.
Here's what happened 🧵
---
The setup:
@ ponsmcp SDK running as an MCP server inside Claude Desktop.
Just two credentials:
- Robinhood Chain RPC API key
- Wallet private key
Claude automatically discovers all 25 available tools on startup.
From there, everything happens through natural language prompts — from querying live chain data to executing transactions.
---
First test: tokenized stock screening.
Asked Claude to scan all 19 verified stock tokens on Robinhood Chain and rank them by liquidity.
Here's what it returned:
META: $4.35M liquidity, +0.69%
NVDA: $2.93M liquidity, +2.07%
TSLA: $860K liquidity, +3.95%
PLTR: $621K liquidity, +21.43%
SNDK: $266K liquidity, +24.28%
Live DEX data. Pulled directly from on-chain trading pairs.
All assets are tokenized debt securities, not equity shares.
---
Second test: Pons launch scanner.
Claude scanned 48 active launches and ranked them based on graduation progress and market activity.
One result stood out in this snapshot: SHROUD CARD.
→ 16% toward the 4.2 ETH graduation threshold
→ +105% 24h change
→ Active buys detected
Then Claude generated a research brief covering utility, deployer information, social links, and DEX liquidity.
One prompt. Multiple tools working together.
---
Final test: an actual on-chain payment.
The execution flow:
`pons_quote → pons_send_token → pons_tx_status`
Result:
✅ 0.01 USDG transferred
✅ Confirmed on-chain
✅ Single-block confirmation
Transaction:
`0xb2f79a554ecba2d2a3188ef4a902943201bad9a73a72deca9a2ad33ba2891ea2`
Gas used: 39,397
The entire workflow — quote, sign, broadcast, and verify — happened through Claude.
No wallet UI interaction required.
---
This is what agent-native infrastructure enables.
Not just an AI that explains what you could do.
An AI agent that can:
- Read live blockchain data
- Analyze token markets
- Discover and research launches
- Execute on-chain payments
- Verify transaction receipts
All through a single MCP integration.
ponsmcp × @AnthropicAI
From natural language to verified on-chain execution.