SUBIT‑Agent Protocol is a structural communication standard built on top of SUBIT‑Lingua v3.0. It defines how agents encode state, intent, actions, plans, and reasoning chains using a deterministic, bit‑exact language of forms, transitions, and sequences.

The protocol provides a minimal, universal interface for agent‑to‑agent and agent‑to‑system communication.

SUBIT‑Agent Protocol uses SUBIT‑Lingua as its transport layer:

• Form — 6‑bit atomic unit
• Word (Transition) — 12‑bit pair of forms
• Sequence — n×6‑bit structural chain

All messages in the protocol are encoded as SUBIT sequences, ensuring full determinism, compactness, and unambiguous parsing.

The protocol defines:

• message types
• packet format
• dialogue patterns
• validation rules
• interpretation rules
• reference implementation

The protocol defines six universal message types:

• STATE — current agent state
• INTENT — desired transition or goal
• ACTION — executable operation
• PLAN — structured multi‑step reasoning
• TRACE — reasoning log
• ERROR — structural or semantic error

Each message carries a SUBIT sequence as its payload.

All communication uses a unified packet format:

{
  "type": "<MESSAGE_TYPE>",
  "payload": "<SUBIT_SEQUENCE>",
  "meta": {
      "agent": "<id>",
      "timestamp": "<iso8601>",
      "session": "<id>"
  }
}

Payloads are always SUBIT‑encoded forms, transitions, or sequences.

The protocol supports deterministic communication flows:

• Request → Response
• State → Intent → Action
• Reasoning Chains
• Multi‑Agent Exchanges
• Error Recovery

These patterns define how agents coordinate, negotiate, and execute tasks.

Validation ensures structural correctness:

• bitstream length must be divisible by 6
• each 6‑bit segment must decode to a valid SUBIT form
• INTENT, ACTION, PLAN require ≥12 bits
• packet type must be valid
• dialogue transitions must follow protocol rules

Agents interpret messages through SUBIT‑Lingua:

• STATE → internal configuration
• INTENT → desired transition
• ACTION → executable step
• PLAN → multi‑step reasoning
• TRACE → explanation
• ERROR → structural failure

Interpretation is deterministic and reversible.

subit-agent-protocol/
│
├── README.md
├── protocol-spec.md
├── architecture.md
│
├── messages.md
├── dialogues.md
├── validation.md
├── interpretation.md
│
├── examples.md
│
└── src/
    ├── encoder.py
    ├── decoder.py
    ├── validator.py
    └── agent.py

The src/ directory includes minimal tools:

• encoder.py — SUBIT encoding
• decoder.py — SUBIT decoding
• validator.py — packet validation
• agent.py — minimal agent implementing the protocol

These components provide a baseline for integrating the protocol into larger systems.

MIT License.