# Sardis Documentation

> Sardis is the Payment OS for the Agent Economy — infrastructure enabling AI agents to make real financial transactions safely through non-custodial MPC wallets with natural language spending policies.

## Documentation Pages

- [A2A Protocol](https://docs.sardis.sh/docs/a2a): Agent-to-Agent (A2A) — Google-developed multi-agent communication protocol for payment coordination, capability discovery, and cross-agent financial messaging.
- [Acceptable Use Policy](https://docs.sardis.sh/docs/acceptable-use): Rules governing the use of Sardis payment infrastructure for AI agents.
- [ACP Protocol](https://docs.sardis.sh/docs/acp): Agentic Commerce Protocol (ACP) — OpenAI's open standard for AI agent commerce. Product feeds, agentic checkout, and delegated payments for ChatGPT-native purchasing.
- [Agent Auth Protocol](https://docs.sardis.sh/docs/agent-auth): How AI agents discover, register, authenticate, and receive spending mandates through the Sardis Agent Auth Protocol. Covers discovery endpoints, capabilities, JWT format, and mandate mapping.
- [AP2 Protocol](https://docs.sardis.sh/docs/ap2): Agent Payment Protocol (AP2) — the Google, PayPal, Mastercard, and Visa consortium standard for secure agent payments. Sardis verifies the full Intent → Cart → Payment mandate chain with signature validation and replay protection.
- [API Reference](https://docs.sardis.sh/docs/api): Complete REST API reference for the Sardis Payment OS. 47+ endpoints covering wallets, payments, mandates, treasury, cards, compliance, and more.
- [Architecture](https://docs.sardis.sh/docs/architecture): Understanding the Sardis system architecture and data flow. How the Policy Engine, MPC wallets, and settlement layer work together.
- [Authentication](https://docs.sardis.sh/docs/authentication): Three ways to start using Sardis. Get your API key via simulation mode, the dashboard, or the CLI.
- [Blockchain Infrastructure](https://docs.sardis.sh/docs/blockchain-infrastructure): How Sardis uses blockchain technology under the hood to provide secure, non-custodial payment infrastructure for AI agents.
- [Changelog](https://docs.sardis.sh/docs/changelog): Release history and version notes for Sardis. Track new features, improvements, security fixes, and bug fixes.
- [Sardis vs Every Alternative](https://docs.sardis.sh/docs/comparison): Detailed comparison of AI agent payment platforms: Sardis vs Stripe, Circle, Fireblocks, Skyfire, Payman AI, and Coinbase.
- [Deployment Guide](https://docs.sardis.sh/docs/deployment): Production-like staging deployment for Sardis API, dashboard, and demo live mode on Cloud Run and AWS App Runner.
- [Error Codes](https://docs.sardis.sh/docs/error-reference): Reference for error codes returned by the Sardis API, with causes and resolution steps for each error type.
- [Frequently Asked Questions](https://docs.sardis.sh/docs/faq): Common questions about Sardis, AI agent payments, spending policies, integrations, security, and pricing.
- [Get Your API Key](https://docs.sardis.sh/docs/get-api-key): Three ways to start using Sardis. Pick the one that fits your workflow -- simulation mode, dashboard signup, CLI wizard, or MCP for Claude.
- [Holds](https://docs.sardis.sh/docs/holds): Pre-authorization holds for reservations and delayed capture. Reserve funds without executing a payment.
- [Overview](https://docs.sardis.sh/docs): Sardis is the Payment OS for the Agent Economy — infrastructure enabling AI agents to make real financial transactions safely.
- [Google ADK Integration](https://docs.sardis.sh/docs/integration-adk): Add payment capabilities to Gemini agents using Google's Agent Development Kit with Sardis FunctionTool instances.
- [Anthropic Agent SDK Integration](https://docs.sardis.sh/docs/integration-agent-sdk): Build Claude-powered agents with native payment tools using Anthropic's Agent SDK. 6 tools as tool_use JSON schemas with automatic execution.
- [AutoGPT Integration](https://docs.sardis.sh/docs/integration-autogpt): Add Sardis payment blocks to AutoGPT workflows with typed Pydantic input/output schemas and policy enforcement.
- [Browser Use Integration](https://docs.sardis.sh/docs/integration-browser-use): Add policy-controlled payments to Browser Use agents with prompt-injection protection, origin allowlisting, and session binding.
- [Composio Integration](https://docs.sardis.sh/docs/integration-composio): Register Sardis payment tools with Composio to give any LLM agent policy-controlled payment capabilities.
- [CrewAI Integration](https://docs.sardis.sh/docs/integration-crewai): Build multi-agent finance teams with shared budgets and payment capabilities using CrewAI and Sardis.
- [LangChain Integration](https://docs.sardis.sh/docs/integration-langchain): Add payment capabilities to your LangChain agents with native tool support. SardisToolkit provides 5 specialized tools for payment operations.
- [MPP & Tempo Integration](https://docs.sardis.sh/docs/integration-mpp): Connect Sardis to the Model Payment Protocol (MPP) and Tempo network for AI-native payments with policy enforcement and audit trails.
- [n8n Integration](https://docs.sardis.sh/docs/integration-n8n): Use the Sardis node in n8n workflows to execute policy-controlled payments from visual automation pipelines.
- [OpenAI Agents SDK Integration](https://docs.sardis.sh/docs/integration-openai): Add payment capabilities to OpenAI Agents SDK agents with native function_tool support. Three tools with automatic schema generation.
- [Vercel AI SDK Integration](https://docs.sardis.sh/docs/integration-vercel-ai): Add payment capabilities to your AI chatbot with the Vercel AI SDK and Sardis. Works with streamText, generateText, and useChat.
- [Framework Integrations](https://docs.sardis.sh/docs/integrations): Sardis works with every major AI agent framework. One payment layer, every platform.
- [MCP Server](https://docs.sardis.sh/docs/mcp-server): 52 tools for Claude, Cursor, and MCP-compatible AI assistants. Zero-code payment integration.
- [Merchant Categories and MCC Codes](https://docs.sardis.sh/docs/merchant-categories): Control which types of merchants your AI agent can transact with using industry-standard MCC codes.
- [Payment Objects](https://docs.sardis.sh/docs/payment-objects): UTXO-inspired one-time payment tokens with a 22-state lifecycle, funding cells, and cryptographic merchant verification.
- [Payments](https://docs.sardis.sh/docs/payments): Execute AI agent payments via bank transfer, virtual card, or stablecoins. Every payment passes policy checks and compliance screening before chain execution.
- [Policy Engine](https://docs.sardis.sh/docs/policy-engine): Every transaction passes through the Sardis Policy Engine before execution. Define spending rules in natural language or structured JSON.
- [Privacy Policy](https://docs.sardis.sh/docs/privacy): How Sardis Labs, Inc. collects, uses, discloses, and safeguards your information.
- [Production Migration Guide](https://docs.sardis.sh/docs/production-guide): Take your agent payments from simulation to mainnet in 3 phases. No code changes required between environments.
- [Protocol Stack](https://docs.sardis.sh/docs/protocols): Industry-standard protocols for secure agent payments. Sardis implements AP2, UCP, A2A, TAP, x402, and ACP for comprehensive agent-to-merchant and agent-to-agent financial operations.
- [Provider Diligence Sheets](https://docs.sardis.sh/docs/provider-diligence): One-page response sheets for Stripe, Lithic, Rain, and Bridge covering funding model, compliance ownership, and operational go/no-go criteria.
- [Quick Start](https://docs.sardis.sh/docs/quickstart): Get Sardis running in under 5 minutes. Install the SDK, create an MPC wallet, set a spending policy, and execute your first AI agent payment.
- [Risk Disclosures](https://docs.sardis.sh/docs/risk-disclosures): Important risk disclosures for using Sardis and digital asset payments. Cryptocurrency, blockchain, custody, AI agent, and regulatory risks.
- [Roadmap](https://docs.sardis.sh/docs/roadmap): Sardis development roadmap from foundation through general availability. Subject to change based on community feedback and market conditions.
- [Runtime Guardrails](https://docs.sardis.sh/docs/runtime-guardrails): Production runtime controls that keep agentic payments deterministic, auditable, and fail-closed.
- [sardis.pay() API](https://docs.sardis.sh/docs/sardis-pay): The primary developer entry point for executing payments through Sardis. Covers explicit payments, auto-routing, error handling, and code examples in Python and TypeScript.
- [SDK Reference](https://docs.sardis.sh/docs/sdk-overview): Complete API reference for Sardis Python and TypeScript SDKs. Payments, wallets, policies, and framework integrations.
- [SDK Reference](https://docs.sardis.sh/docs/sdk-reference): Official Python and TypeScript SDKs for Sardis. Typed resources and validated models for wallets, payments, cards, policies, and more.
- [Security Model](https://docs.sardis.sh/docs/security): Sardis security architecture: non-custodial MPC wallets via Turnkey, 12-check policy pipeline, KYC via Didit, AML via Elliptic, Travel Rule compliance, and virtual card ASA real-time screening.
- [Spending Mandates](https://docs.sardis.sh/docs/spending-mandates): The authorization primitive that makes agent payments safe. Machine-readable payment permissions with scoped, time-limited, revocable authority.
- [TAP Protocol](https://docs.sardis.sh/docs/tap): Trusted Agent Protocol (TAP) — cryptographic identity verification for AI agents with Ed25519/ECDSA signatures, attestations, and trust scoring.
- [Terms of Service](https://docs.sardis.sh/docs/terms): Terms of Service for Sardis Labs, Inc. payment infrastructure for AI agents.
- [Time-Based Policies](https://docs.sardis.sh/docs/time-based-policies): Restrict agent transactions to specific hours, days, and timezones. Control when your AI agent can spend money.
- [Troubleshooting Guide](https://docs.sardis.sh/docs/troubleshooting): Diagnose and fix common Sardis integration issues including wallet configuration, policy blocks, authentication failures, chain timeouts, rate limits, and debug logging.
- [Trust Center](https://docs.sardis.sh/docs/trust-center): How Sardis protects your agents, your money, and your data. Security architecture, compliance status, subprocessors, SLA, and data processing.
- [UCP Protocol](https://docs.sardis.sh/docs/ucp): Universal Commerce Protocol (UCP) — standardized checkout flow for AI agents with cart management, discount application, session tracking, and fulfillment.
- [Wallets](https://docs.sardis.sh/docs/wallets): Non-custodial MPC wallets for AI agents. Create, fund, and manage wallets with spending policies across multiple chains.
- [Webhooks](https://docs.sardis.sh/docs/webhooks): Receive real-time notifications when events occur in your Sardis account. Signature verification, retry behavior, and event types.
- [Sardis Whitepaper](https://docs.sardis.sh/docs/whitepaper): The Payment OS for the Agent Economy: Preventing Financial Hallucinations Through Programmable Trust.
- [Who Owns Accountability When an AI Agent Moves Money?](https://docs.sardis.sh/docs/blog/agent-accountability): When an AI agent pays the wrong vendor, overspends, or gets exploited -- who is responsible? The operator, the model provider, the framework, or the agent itself?
- [Fiat Rails: Bridging Traditional Banking to Agent Wallets](https://docs.sardis.sh/docs/blog/fiat-rails): Fiat Rails is the Sardis design-partner lane for bank funding, card spend, and payout orchestration with policy enforcement. Stablecoins remain available as an optional settlement rail.
- [Financial Hallucination Prevention: Why AI Needs Guardrails](https://docs.sardis.sh/docs/blog/financial-hallucination): AI agents can hallucinate financial transactions just like they hallucinate facts. How cryptographic policy enforcement prevents unauthorized spending before funds move.
- [Blog](https://docs.sardis.sh/docs/blog): Updates, tutorials, and deep dives from the Sardis team on AI agent payments infrastructure.
- [Introducing Sardis: Secure Payments for AI Agents](https://docs.sardis.sh/docs/blog/introducing-sardis): Sardis provides MPC wallets and natural language policy enforcement so AI agents can transact autonomously while preventing financial hallucination errors.
- [MCP Server: 52 Tools for AI Payments](https://docs.sardis.sh/docs/blog/mcp-52-tools): Our MCP server has grown from 4 tools to 52. From treasury ACH rails to virtual cards, learn how to give Claude or Cursor complete payment capabilities.
- [MCP Integration: Zero-Code AI Payments in Claude](https://docs.sardis.sh/docs/blog/mcp-integration): Add Sardis payment capabilities to Claude Desktop in under 5 minutes using the MCP server. No code required. 52 tools for payments, wallets, treasury, and checkout.
- [Understanding MPC Wallets for Agent Security](https://docs.sardis.sh/docs/blog/mpc-wallets): Multi-Party Computation wallets distribute key shares so no single entity can move funds. Learn how Sardis uses threshold ECDSA to provide non-custodial, policy-enforced wallets for AI agents.
- [Policy Engine Deep Dive: Configuring Spending Rules](https://docs.sardis.sh/docs/blog/policy-engine): Explore the full capabilities of the Sardis policy engine. From simple spending limits to vendor allowlists, time-based rules, velocity controls, and approval workflows.
- [Sardis AI Agent Payments: What It Is and How It Works](https://docs.sardis.sh/docs/blog/sardis-ai-agent-payments): A practical guide to Sardis AI agent payments infrastructure -- deterministic policy enforcement, approvals, cards, and on-chain rails.
- [SDK v0.2.0: LangChain, OpenAI, and LlamaIndex Support](https://docs.sardis.sh/docs/blog/sdk-release): SDK v0.2.0 adds native integrations for LangChain, OpenAI function calling, and LlamaIndex, plus batch payments, transaction webhooks, and improved TypeScript types.
- [How AI Agent Spending Rules Actually Work](https://docs.sardis.sh/docs/blog/spending-rules-explained): No jargon, no code. A practical guide to how Sardis spending policies control what AI agents can and cannot do with money.
- [Understanding AP2: The Industry Standard for Agent Payments](https://docs.sardis.sh/docs/blog/understanding-ap2): AP2 (Agent Payment Protocol) is the Google, PayPal, Mastercard, and Visa consortium standard for AI agent transactions. Learn how the mandate chain provides cryptographic proof of authorization.
- [Sardis v0.5: UCP and A2A Protocol Support](https://docs.sardis.sh/docs/blog/v05-protocols): Sardis v0.5 adds full support for UCP (Universal Commerce Protocol) and A2A (Agent-to-Agent) protocol, enabling standardized checkout flows and multi-agent communication.
- [Sardis v0.7: Production Hardening and Fireblocks Integration](https://docs.sardis.sh/docs/blog/v07-production-hardening): Sardis v0.7 eliminates 24 technical debt items, migrates to PostgreSQL persistence, adds Fireblocks MPC custody, and hardens authentication across every API route.
- [Sardis v0.8.1: Protocol Conformance Hardening for AP2/TAP](https://docs.sardis.sh/docs/blog/v081-protocol-conformance): v0.8.1 tightens AP2 payment semantics, strengthens TAP validation paths, and adds a source-mapped conformance baseline for canonical protocol references.
- [Sardis v0.8.2: Release Readiness Hardening for MCP + SDKs](https://docs.sardis.sh/docs/blog/v082-release-readiness): v0.8.2 hardens release scripts, isolates protocol conformance lanes, and improves local validation when npm or network access is constrained.
- [Sardis v0.8.3: Demo Ops + Cloud Deployment (Cloud Run & AWS)](https://docs.sardis.sh/docs/blog/v083-demo-ops): v0.8.3 adds deployment automation for Google Cloud Run and AWS App Runner, tightens live-mode operator UX, and documents the frontend-to-backend integration contract.
- [Sardis v0.8.4: Packages Live on npm & PyPI + Security Audit](https://docs.sardis.sh/docs/blog/v084-packages-live): All 19 Sardis packages are now published to public registries. This release also includes a comprehensive security audit with 54 fixes across 8 batches.
- [Sardis v0.8.8: ERC-4337 Base Preview + Truth Alignment](https://docs.sardis.sh/docs/blog/v088-erc4337-preview): v0.8.8 adds a real ERC-4337 implementation lane for design partners on Base Sepolia, introduces account_type parity across API and SDKs, and tightens public language around non-custodial posture.
- [Sardis v0.9.0: Multi-Provider Fiat Rails + AI Framework Integrations](https://docs.sardis.sh/docs/blog/v09-multi-provider-fiat): v0.9.0 ships Stripe Treasury + Issuing for fiat operations, Coinbase Onramp for zero-fee USDC purchases, a sub-ledger system, and integrations across every major AI framework.
- [Sardis v0.9.5: Strict Live Mode + Operations Hardening](https://docs.sardis.sh/docs/blog/v095-strict-live-ops): v0.9.5 ships strict live-mode controls, replay/idempotency proof gates, SLO alerting, DR evidence automation, and runtime security-policy preflight visibility.
- [Why Sardis: The Policy Firewall for Agent Payments](https://docs.sardis.sh/docs/blog/why-sardis): Sardis fills a critical gap in the agent payment landscape with natural language policy enforcement, non-custodial MPC security, virtual cards, and zero-config MCP server integration.

## Full Content

For the full documentation content suitable for LLM consumption, see:
- [llms-full.txt](https://docs.sardis.sh/llms-full.txt)