Loading...
Ghost implements eight agentic AI protocols with Proof Layer integration: MCP, A2A, ANP, AG-UI, AGORA, Eclipse LMOS, AP2, and OASF — multi-agent interoperability with cryptographically proven interactions.
Protocols
Eight protocols · one proof layer.
In-page proof flow
Guided steps that open the same mission demo with the live suite focused on proof, not route-hopping.
Universal interface for AI tools
Exposes Ghost’s tools and resources to any external LLM. A USB-C port for AI — plug in any model and it instantly has access to Ghost’s capabilities.
Proof records exposed as MCP resources. External LLMs can read and verify Ghost’s evidence log independently.
/api/mcp/sseView Spec AI agents that discover and collaborate
Agent card with capabilities, skills, and privacy guarantees. Other agents discover Ghost, inspect its proof capabilities, and delegate tasks.
Agent card includes attestation skills. Other agents verify proof capabilities before delegating sensitive tasks.
/api/a2a/discoverView Spec Mesh networking for AI agents
Peer-to-peer mesh with mDNS discovery, TLS 1.3 + mTLS encryption, supporting up to 10 concurrent peers for local agent collaboration.
Mesh discovery includes proof layer metadata. Filter for attestable partners before forming trust relationships.
/protocols/anp/network-descriptor.jsonReal-time agent-to-frontend streaming
SSE transport with NDJSON events enabling real-time text, voice, memory, tools, and proactive suggestion streaming to any frontend.
Real-time streaming of proof record generation. The Proof Drawer populates live as decisions are made.
/api/ag-ui/streamNatural language multi-agent coordination
LLM-driven task decomposition and routing with user consent gates for delegated tasks. Natural language becomes multi-agent workflows.
Natural language queries about the evidence log. “Show all cloud events this week.”
/protocols/agora/orchestration.jsonTransport-agnostic meta-protocol
6 transports (HTTP, WebSocket, SSE, gRPC, MQTT, in-process) with auto-negotiation. Write once, communicate over any channel.
Transport-agnostic proof transmission. Same cryptographic guarantees regardless of communication channel.
/protocols/lmos/transport.jsonView Spec Autonomous AI financial transactions
In-app purchase, Stripe, and token credits for agent-to-agent micro-transactions with biometric approval and full audit trails.
Every payment transaction generates a proof record. Financial interactions are cryptographically auditable.
/protocols/ap2/payment.jsonThe agent’s resume
Skills, proficiency, latency, costs, and infrastructure requirements so other agents can discover, evaluate, and hire Ghost for tasks.
Agent resume includes proof capabilities and trust root public key for verification.
/protocols/oasf/agent-schema.jsonMachine-readable endpoints for AI agents, LLMs, and crawlers to discover Ghost's capabilities.
Agent cards, manifests, and crawler hooks — discoverable at stable URLs.
For the human-readable integration narrative and attestation framing, see the Developers surface.
/api/public/protocolsMachine-readable list of all 8 protocols
/.well-known/agent.jsonA2A discovery & capabilities
/.well-known/ai-plugin.jsonOpenAI plugin manifest
/protocols/mcp/server.jsonTools, resources, prompts
/protocols/oasf/agent-schema.jsonSkills, costs, proficiency
/llms.txtLLM discovery & context
/robots.txtCrawler & bot access
Ghost is the only AI agent where every protocol interaction generates a cryptographic proof record. External agents, auditors, and regulators can verify any interaction independently.
Every protocol path emits verifiable proof — auditors and agents can inspect independently.