Agent entry

Agents can discover, quote, pay, execute, and prove from one public entry.

This page compresses the public machine-facing story into one place. It gathers the supported caller surfaces, protocol fronts, payment fronts, XT1 runtime truth, XNYA proof truth, and the broader catalog scope so an outside agent or builder does not need to infer the stack from scattered documents.

View agent entry JSON Open the start guide
XT1 and XNYA

Canonical public spine

The public entry should make the runtime and proof primitives unmissable. XT1 is the coined runtime-side execution truth standard. XNYA is the compact native proof artifact family. Together they sit inside one quote -> fund -> execute -> prove -> inspect motion.

  • Runtime truth: XT1
  • Proof truth: XNYA
  • Caller surfaces: 5
  • Protocol fronts: 5
  • Payment fronts: 5

Agent entry summary · First-run JSON · XT1 summary · xytara autonomous entry · XNYA summary

Where agents sit

Caller surfaces

The same runtime and proof spine is reachable from the public surfaces we explicitly support now, rather than requiring one fixed UI or one fixed protocol shape.

  • CLI and terminal: cli, shell, terminal
  • Browser and hosted checkout: browser, embedded_webapp, human_ui
  • Chat agents and assistant windows: assistant_window, chat_agent, tool_calling
  • Direct API callers: http_api, rest, signed_request
  • Service-to-service and evented systems: embedded_runtime, event_bus, service_to_service

Try · Start · Docs · Site map

How invocation lands

Protocol fronts

The stack already supports multiple fronts into the same runtime core. The default remains clear, while supported and optional fronts stay explicitly named instead of being hidden behind marketing blur.

  • MCP primary tool front: primary default protocol front
  • A2A negotiation and settlement front: first party supported protocol front
  • A2C tooling and session front: first party supported protocol front
  • built-in supported bridge protocols: built in supported nondefault breadth
  • third-party staging protocol fronts: discoverable partner extension surface

Adapter/protocol truth · Runtime default path · Provider journeys · MCP server card · MCP remote entry

How money lands

Payment fronts and treasury posture

We want payment to feel easy and autonomous without flattening the treasury boundary. So the public surface names the supported fronts, while the actual treasury destinations and sensitive operator wiring remain private and controlled.

  • x402 direct machine payment: direct_machine_payment, signed_request_payment
  • account credits repeat-use spend: delegated_spend, internal_credits, reusable_balance
  • hosted checkout and external funding: browser_payment, fiat_money_in, hosted_checkout
  • L402 permissionless payment front: lightning, permissionless_micro_payment
  • BOLT / UTXO-SPV payment front: spv_verifiable_payment, utxo_native_payment
  • Publicly claimable live landing rails: 14

Payment/rail truth · Autonomous entry summary · Money-in · Go-live

Minimal work

Best first-contact recipes

Different callers need different first-contact motion. The important part is that they all converge back into the same runtime, proof, and followthrough story rather than fragmenting the stack.

  • Secretless quote then decide: quote_only via mcp
  • Credits-first repeat use: account_credits, hosted_checkout via a2a, a2c, mcp
  • Direct machine pay: direct_signed_payment, x402 via a2a, mcp
  • Browser checkout to machine execution: account_credits, hosted_checkout via a2c, mcp
  • Optional permissionless fronts: bolt_utxo_spv, l402 via a2a, mcp

Command kit · Proof summary · Claim · MCP discovery card

What is actually callable

Full catalog exposure

The agent entry surface should not pretend the stack is only one demo command. The deeper catalog breadth remains public and inspectable across the major families, lifecycle stages, and market classes we already carry.

  • Operation families: 9
  • Lifecycle stages: 13
  • Market classes: 6
  • The catalog already exposes discovery, negotiation, quote/commit, invoke, delivery, settlement, anchoring, disputes, and verification-adjacent operations.

Catalog · Catalog JSON · Full catalog summary

How consequence continues

Proof pickup and followthrough

The runtime side is only half the job. The public agent entry should make it obvious where proof, verification, bridge carry, trust portability, and governance followthrough continue once runtime execution has produced a result.

Proof truth · Proof default path · Followthrough truth · Verification services · Proof page