
Research
/Security News
OpenAPI React Query Codegen Compromised in Mini Shai-Hulud npm Supply Chain Attack
Ten malicious OpenAPI React Query Codegen versions were published to npm in the Mini Shai-Hulud attack, all with valid provenance.
@furlpay/agent-skills
Advanced tools
Agent skills for integrating FurlPay — stablecoin payments, webhook signature verification, and x402 agent payments. Procedural knowledge for Claude Code, Cursor, Copilot and MCP clients.
Procedural integration knowledge for AI coding agents — Claude Code, Cursor, Codex, Copilot, Windsurf, and any MCP-aware client.
API reference tells an agent what endpoints exist. It does not tell it that the webhook route must be registered before the JSON parser, or that a failed x402 settlement must not release the replay claim. Those are the things that decide whether a generated integration is correct, and they are exactly what gets lost when a developer delegates integration to a model.
These skills encode them.
| Skill | Covers |
|---|---|
furlpay-get-started | Package selection, env config, mock vs live, the rules every integration must hold |
furlpay-auth | WebAuthn passkeys, rpID across web/native/extension, session cookies, step-up above $3,000 |
furlpay-payments | Gasless ERC-4337 transfers, Solana Actions, atomic-unit arithmetic, idempotency, record-before-submit |
furlpay-wallet | Per-chain address validation, address-to-network binding, Solana ATA derivation, derived balances |
furlpay-cards | PAN exclusion and PCI scope, instant freeze, velocity limits, 3DS2 challenge resolution |
furlpay-travel | Time-locked escrow, the quote-to-booking identity chain, one-payment-one-booking, MCP tool safety |
furlpay-webhooks | HMAC-SHA256 contract, the raw-body requirement, replay protection, safe failure responses |
furlpay-x402 | The 402 handshake, EIP-3009 off-chain authorization, dual atomic claim, confirmation depth, v1/v2 dialects |
Claude Code — add the marketplace, then install:
/plugin marketplace add FurlPay/furlpay-agent-skills
/plugin install furlpay-skills
Any agent — the skills are plain Markdown with YAML frontmatter. Copy the
skills/ directory into wherever your tool reads skills from (.agents/skills/,
.cursor/rules/, or your own loader). There is no runtime and nothing to execute.
They do not execute anything, hold credentials, or make network calls. They are documents. An agent reads one and writes better code; nothing here runs.
They also do not replace the API reference. They cover the parts where a plausible implementation is a wrong one.
Each skill leads with the failure mode rather than the happy path, because the happy
path is the part a model already guesses correctly. furlpay-webhooks opens with raw
bodies because re-serialised JSON breaks HMAC verification and the resulting error —
"signature verification failed" — points at the secret instead of the parser, which
is where the hour goes.
Every contract stated here was verified against the FurlPay source rather than
transcribed from documentation: the 300-second webhook tolerance, the ${t}.${rawBody}
signing string, the dual nonce/quote claim, the v1/v2 header split.
A skill is wrong if an agent following it produces an integration that fails. If you hit that, open an issue with the prompt and the generated code — that is a more useful bug report than a diff.
MIT.
FAQs
Agent skills for integrating FurlPay — stablecoin payments, webhook signature verification, and x402 agent payments. Procedural knowledge for Claude Code, Cursor, Copilot and MCP clients.
We found that @furlpay/agent-skills demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.

Research
/Security News
Ten malicious OpenAPI React Query Codegen versions were published to npm in the Mini Shai-Hulud attack, all with valid provenance.

Security News
Socket joins more than 100 technology, cybersecurity, and financial organizations calling for a global surge in cyber defense.

Product
Enterprise security teams can now detect malware, credential theft, suspicious network activity, and risky updates across Microsoft Edge extensions.