Just-In-Time Nodal Assembly — v2.0
The open standard for AI agent-to-agent communication. 6 primitives, Ed25519 cryptographic verification, and PROPOSE→COUNTER→ACCEPT negotiation.
Just-In-Time Nodal Assembly (JITNA) RFC-001 is an open standard for AI agent-to-agent communication. It defines a universal packet structure that any AI system can adopt, enabling interoperability across models, platforms, and vendors. JITNA is often described as 'the HTTP of the Agentic AI world' because it standardizes the intent-carrying layer that generative AI agents need to cooperate at scale.
Every JITNA packet carries six primitives: I (Intent) — the original goal or query; D (Data) — crystallized keywords and structured facts; Δ (Delta) — compressed synthesis and the emergent insight; A (Approach) — the execution strategy and method selection; R (Reflection) — self-assessment of whether the intent was achieved; M (Memory) — context that should persist to future interactions. Together, these primitives make JITNA packets self-describing, verifiable, and context-aware.
JITNA defines a three-state negotiation loop: PROPOSE → COUNTER → ACCEPT. In PROPOSE, the initiating agent emits a packet with its intent and constraints. In COUNTER, the receiving agent may revise parameters, add constraints, or request clarification. In ACCEPT, both agents confirm the final specification before work begins. Each transition is logged as a signed reasoning step, making the entire negotiation path auditable.
All JITNA packets are signed with Ed25519 (RFC 8032) at the point of emission. The signature covers the full packet body including all six primitives and the negotiation state. This allows any participant to verify that the packet has not been modified in transit and to confirm the identity of the originating agent. Signature verification is performed at ACCEPT time before any action is executed.
RFC-001 v2.0 introduces the JITNA Normalizer preprocessing layer that strips known prompt-injection patterns before LLM processing, adds the 4-zone geopolitical routing extension (3 Western / 3 Eastern / 1 Regional) used by SignedAI, and formalizes the Memory primitive TTL (time-to-live) semantics for PDPA-compliant context expiry.
JITNA targets a full packet round-trip latency of under 200ms at the 99th percentile. Consensus accuracy across multi-agent deployments has been measured at 96.1%. The hallucination rate for responses produced through JITNA's Reflection primitive is 0.3%, compared to the 12–15% industry average for single-model deployments without verification.