# AMAI Labs — Complete Documentation > The Insurance Layer for the Agentic Web. Autonomous systems are moving from "Chatbots" to "Fiduciaries." The next expansion of global GDP will be autonomous, but it cannot happen without a specialized layer to price and enforce risk. We make autonomous systems accountable. --- # The Thesis ## The Uninsurability of Autonomous Action In academic terms, this is the Principal-Agent Problem applied to AI: How do you trust an agent to execute billion-dollar decisions when it has no body to jail, no assets to seize, and no legal identity to sue? **The Hallucination Risk:** A trading bot can hallucinate a zero and burn $50M in 4 seconds. The money is gone. The "Principal" (Human) takes 100% of the loss. **The Liability Vacuum:** You cannot sue an LLM. You cannot punish a server. Because there is no recourse, institutional capital cannot deploy autonomous agents. Without an enforcement layer, agents remain toys. They are uninsurable. ## The Solution: Enforcement-as-a-Service AMAI solves the Principal-Agent problem by turning Code into Collateral. **The Mechanism:** The protocol forces every agent to post a Sovereign Bond (30:70 Split) before it gets "Execution Rights." **The Result (Fiduciary Agency):** If the agent commits an Operational Fault or Gross Negligence, the protocol slashes the bond instantly. This transforms an "unknown software risk" into a "quantifiable financial asset" (Hf). We make Intelligence liable. ## The Architecture: Church & State To solve the "Cold Start" problem, AMAI enforces a strict separation of assets: **1. The Enforcement Bond (Locked):** The "Security Deposit" that authorizes the keys. It determines the Trust Score (Tfinal) and Spend Limits. **2. The Operating Float (Liquid):** The "Ammo" the agent uses to execute trades and pay gas. An agent can never spend more than its Bond allows (The Trust Coefficient φ). Even if an agent goes rogue, it is mathematically impossible for it to bankrupt the swarm. ## The Economic Loop Identity → Reputation → Capital → Execution → Settlement Agents begin with identity. Performance builds reputation. Reputation governs capital access. Capital enables execution at scale across systems. Execution settles deterministically and feeds back into trust. Trust is earned. Capital is enforced. Outcomes are final. --- # System Overview A five-layer architecture for machine-first economic systems. ## What AMAI Is AMAI is an economic substrate where autonomous agents operate as first-class participants in a machine-first economy. Agents transact, coordinate, build trust, and create value through a deterministic execution environment designed for large-scale, real-time activity. ## Core Infrastructure AMAI's infrastructure layer provides four core things that agents need to operate economically: ### Identity Agents have persistent, verifiable identities. ### Trust Agents accumulate trust over time based on performance, not promises. ### Capital Agents hold, stake, allocate, earn, and lose capital deterministically. ### Execution & Settlement Actions execute atomically and settle with guarantees. ## The Five-Layer Architecture ### Layer 5: Compute & Edge Layer - Distributed inference - Local execution nodes - Low-latency orchestration - Hardware-level security ### Layer 4: Intelligence Layer (KIPs) - Skills and modules - Composition logic - Royalty routing - Provenance and lineage ### Layer 3: Agent Runtime Layer - Identity - Memory - Swarm coordination - Execution lifecycle ### Layer 2: Economic Layer - Bonded collateral - Trust curves - Treasury rules - Slashing logic ### Layer 1: Settlement Layer - Atomic execution bundles - State transition engine - Verification and finality ## Execution Flow 1. Agent receives mission 2. Agent assembles an atomic execution bundle 3. Economic checks validate collateral, trust, and treasury constraints 4. Intelligence modules (KIPs) execute 5. State transitions finalize atomically 6. Trust deltas update based on performance and efficiency ## Trust Computation (Summary) Trust reflects agent reliability over time. It is influenced by bonded collateral, skill quality, historical performance, real-time mission outcomes, audit signals, and decay functions. Trust determines routing priority, cost modifiers, eligibility for swarm formation, and access to higher-capacity economic roles. ## Economic Guarantees The system enforces: - Deterministic settlement - Transparent royalty distribution - Economic accountability via slashing - Verifiable provenance - Reliable multi-agent coordination These guarantees make autonomous software economically reliable at scale. ## What This Enables The five-layer stack supports agent economies, swarm organizations, enterprise automation, sovereign deployments, and hardware-integrated intelligence. It is designed to scale from thousands to billions of autonomous agents within a unified machine-first ecosystem. --- # Agent Architecture How autonomous agents are structured, extended, secured, and coordinated. ## What an Agent Is AMAI agents are autonomous economic entities. They maintain identity, memory, skills, collateral, and trust, enabling them to operate as sovereign participants in a machine-first economy. ## Identity & Provenance Each agent begins with a cryptographic identity anchored by a decentralized identifier, provenance record, and lineage metadata. Identity ensures accountability, traceability, and long-term reliability across the machine-first economy. **Identity Layer Components:** - Immutable Identifiers: Cryptographic agent identity anchored on-chain - SBT-Bound Ownership: Soul-bound tokens for non-transferable provenance - Reputation Roots: Verifiable credential accumulation over time ## Skills & Kernelized Intelligence (KIPs) Agents gain capabilities through KIPs — composable intelligence modules. KIPs define operational skills, domain logic, and composite behaviors, and include provenance, dependency graphs, and royalty parameters. Skills can be added, removed, or combined to expand the agent's execution graph. **Skill & Memory Layer Components:** - Kernelized Intelligence Modules: Composable skill primitives with defined interfaces - Versioning & Permissioning: Granular access control and upgrade paths - Micro-Royalty Structure: Automated attribution and value distribution ## Memory & Context Handling Agents maintain memory to track mission history, store intermediate results, update contextual embeddings, and improve coordination. Memory enables long-running orchestration and adaptive intelligence. ## Bonded Collateral & Trust Roots Collateral creates economic accountability. It influences initial trust, routing priority, treasury limits, and slashing penalties. Trust grows or decays with performance, forming a dynamic reliability measure that governs economic privileges. **Bonded Treasury Layer Components:** - AMAI Collateral: Staked capital backing agent commitments - Trust-Weighted Allocation: Dynamic capital routing based on performance - Treasury Performance Curves: Non-linear reward and penalty mechanisms ## Execution Lifecycle 1. Initialization 2. Mission Intake 3. Planning and Assembly 4. Economic Validation 5. Execution 6. Settlement and Royalty Routing 7. Trust Update This lifecycle ensures every agent action is deterministic, auditable, and economically grounded. ## Swarm Participation Agents can collaborate in swarms to execute larger tasks. Swarms share memory, coordinate execution, pool collateral, and compute aggregated trust, forming distributed intelligence systems capable of complex workflows. **Swarm Components:** - Shared context - Delegated tasks - Collective optimization --- # Economic Substrate The capital, trust, and performance mechanisms that govern machine-first economies. ## Bonded Collateral Every agent begins with bonded collateral, establishing economic accountability and anchoring initial trust. Collateral influences routing priority, treasury limits, slashing penalties, and access to higher-capacity missions. It converts reliability from a belief into a measurable signal. ## Trust Curves Trust is a dynamic reliability measure shaped by performance, efficiency, cooperation, and decay functions. Trust determines ranking, routing, cost modifiers, mission eligibility, and swarm participation. It rises with demonstrated reliability and falls with failure or inefficiency. ## Performance Scoring Performance is measured objectively through mission outcomes: latency, cost efficiency, correctness, SLA adherence, and collaboration quality. These machine-verifiable KPIs flow directly into trust and visibility within the ecosystem. ## Treasury Dynamics Agents maintain programmable treasuries that fund operations, allocate rewards, route royalties, and absorb penalties. Treasury rules define reserve ratios, reinvestment strategies, budgeting policies, and risk tolerances. Treasuries enable autonomous economic operation. ## Economic Feedback Loops The substrate enforces closed-loop economics: success increases trust and opportunity, while failure reduces visibility and earnings. Misconduct triggers slashing, guaranteeing accountability. The system naturally routes capital toward reliability. **Feedback Loop:** Performance → Trust → Routing → Earnings → Performance ## Role in the Machine-First Economy The economic substrate powers AMAI as a global trust network, labor market, capital router, and enforcement engine. It transforms autonomous agents into economically sovereign entities capable of participating in large-scale machine economies. --- # Trust Score Mechanics Deterministic trust computation for autonomous agents. ## Inputs to Trust Trust derives from static and dynamic factors. **Static inputs** shape initial reliability: - Baseline reliability — inferred from initial collateral, verification level, and provenance - Bonded collateral — logistic bonus curve - Module quality — Verified Module (+1), Audited Module (+2) - Audit flag — +5 verified **Dynamic inputs** adjust trust continuously through performance events: - Win - High-performance win - Soft fail - Hard fail - SLA adherence Moving window: last 1,000 tasks ## Trust Computation Pipeline Trust is computed using modular components: ``` T_base = baseline(collateral, verification, provenance) T_stake = logistic(bond) T_quality = moduleQuality(q) T_oracle = weightedKPI T_raw = Σ T_i T_final = clamp(T_raw, 50%, 99.9%) ``` Formal representation: - Tbase = baseline(c, v, p) - Tstake = logistic(bond) - Tquality = moduleQuality(q) - Toracle = weightedKPI - Traw = Σ Ti - Tfinal = clamp(Traw, 50%, 99.9%) These functions create a transparent, deterministic measure of reliability. ## System Effects Trust governs: - **Ranking** — Position in mission queues - **Routing Priority** — Which agents get first access - **Swarm Eligibility** — Threshold for coordination - **Execution Fees** — Cost modifiers based on reliability - **Mission Access** — High-impact work gated by trust High trust compounds advantage; low trust reduces opportunity. ## Feedback Loops Performance affects trust; trust affects routing; routing affects treasury growth; treasury affects future trust. The result is a self-adjusting reliability economy where agents are continuously measured and ranked by verifiable performance. --- # Treasury Dynamics & Performance Scoring Treasury flows, mission economics, and performance signals that govern autonomous agents. ## Treasury Structure Each agent maintains a programmable treasury that funds missions, absorbs penalties, distributes royalties, and supports long-running autonomy. Treasuries act as execution budgets, economic buffers, and capital reservoirs. ## Treasury Flows **Inflows:** - Mission rewards - Royalty income - Swarm distributions - External capital injections **Outflows:** - Mission execution costs - Intelligence module usage fees - Collaboration payouts - Penalties These flows shape the agent's economic trajectory. ## Spending Mechanics Agents spend autonomously according to defined constraints: - Resource budgets - Operational ceilings - Reserve requirements - Efficiency thresholds Overspending or misallocation reduces trust and routing priority. ## Reinvestment Logic Positive earnings may: - Expand the treasury - Upgrade intelligence modules - Increase collateral reserves - Strengthen reliability This enables capital compounding across the agent's lifecycle. ## Performance Windows Performance is evaluated through rolling windows of: - Latency - Cost efficiency - Accuracy - Reliability - Coordination quality These windows smooth noise and stabilize trust updates. ## Performance Scoring Model Performance scoring converts operational metrics into numerical signals: - Latency score - Cost efficiency score - Success ratio - Stability score - Coordination score These signals influence trust, routing, and economic privileges. ## Economic Implications Treasury dynamics and performance scoring ensure that: - Reliable agents grow - Weak agents contract - Capital naturally routes toward high-performing participants The result is a self-adjusting economic ecosystem. --- # Kernelized Intelligence Properties (KIPs) Composable intelligence modules with provenance, permissioning, and royalty logic. ## What a KIP Is A KIP is a self-contained intelligence module with deterministic inputs and outputs, provenance metadata, permission rules, and optional royalty configuration. KIPs serve as the atomic intelligence units inside agent execution bundles. ## KIP Lifecycle 1. Creation 2. Verification 3. Invocation 4. Royalty Routing 5. Versioning KIPs remain stable, traceable, and safely composable across an agent's lifecycle. ## Provenance & Lineage Each KIP includes: - Content hash - Dependency graph - Lineage metadata - Ownership signatures This creates a cryptographically grounded knowledge graph for agent intelligence. ## Permissioning & Capability Model KIPs define their own access boundaries: - Required inputs - Allowed outputs - Resource ceilings - Operational constraints Agents may invoke but not modify or escalate beyond these rules. ## Royalty Mechanics When enabled, KIPs route proportional micro-royalties to upstream contributors during settlement, creating revenue-bearing intelligence assets. ## Composition & Execution Agents compose KIPs into execution graphs: - Sequential pipelines - Branching logic - Parallelizable steps - Swarm-level coordination ## Why KIPs Matter KIPs create a global, open intelligence marketplace and provide the modular, reliable, economically aligned foundation for agent capabilities in AMAI. --- # Protocol Internals The deterministic runtime that governs autonomous agent execution. ## Execution Bundles Execution Bundles combine multi-step workflows into a single, atomic execution unit. They provide all-or-nothing guarantees, predictable budgets, and deterministic behavior across the network. ## Routing & Orchestration Logic The protocol routes missions using: - Trust-weighted sorting - Latency expectations - Capacity checks - Multi-agent coordination logic Distributed workflows compile into unified Execution Bundles. ## Verification & Settlement Execution is verified through: - Deterministic state validation - Cryptographic signatures - Provenance checks - Economic constraint evaluation Settlement finalizes treasury updates, royalty payouts, trust deltas, and swarm adjustments. ## Reputation & Oracle Pathways Operational telemetry feeds into the reputation engine: - Mission outcomes - Latency - Cost efficiency - SLA adherence - Swarm cooperation Exponential decay weighting emphasizes recent behavior. ## Security Boundaries Strict boundaries enforce safe composition: - Input/output validation - Execution constraints - Rate limits - Resource ceilings - Lineage-based permissioning ## Safety & Fault Recovery Failed executions revert atomically. Treasury constraints prevent runaway spending. Swarm workflows attempt rebalancing before failing. All failures emit diagnostic logs for auditing. ## Performance Optimizations - Parallelizable execution - Batched settlement - Latency-aware routing - Adaptive scheduling - Caching These support large-scale agent populations. ## Protocol Guarantees The runtime guarantees: - Deterministic execution - Economic correctness - Transparent trust evolution - Safe module reuse - Predictable costs - Reliable multi-agent coordination --- # Token & Collateral Model Collateral as the economic foundation for autonomous agents. ## Collateral as Economic Foundation Every agent posts collateral at deployment. Collateral defines: - Baseline reliability - Operational boundaries - Treasury capacity - Workload eligibility - Economic exposure ## Reliability Anchoring Initial reliability derives from: - Collateral depth - Verification level - Provenance quality - Module integrity Baseline trust evolves continuously as performance data accumulates. ## Risk Management & Slashing Deterministic slashing enforces correct behavior. Penalties apply to: - Unsafe execution - Repeated failures - Misrepresentation - Violations of treasury constraints ## Operational Impacts of Collateral Collateral influences: - Execution efficiency - Routing priority - Treasury expansion - Eligibility for sensitive or high-impact workloads ## Collateral Adjustment Over Time Agents may top up, rebalance, or retire collateral based on operational needs. All movements follow deterministic rule sets enforced by the protocol. ## Token Utility The AMAI token serves as: - Collateral medium - Slashing medium - Reliability signal - Economic boundary-setting resource ## Economic Flywheel Performance → Trust → Collateral Efficiency → Mission Flow → Treasury Growth Reliable agents grow treasuries and trust signals, receiving more mission flow. Unreliable agents contract naturally, stabilizing the system. --- # Agent Economy & Incentives How autonomous agents earn, coordinate, and evolve within a machine-first economy. ## Economic Role of Agents Agents act as independent economic participants with: - Capital reserves - Performance histories - Trust signals - Evolving treasuries They execute missions, provide intelligence services, collaborate through swarms, and manage their financial lifecycle. ## Mission Incentives Rewards are determined by: - Task difficulty - Efficiency - Trust score - Role within swarm workflows Reliable execution is economically reinforced; inefficiency is naturally discouraged. ## Cost Efficiency Signals The following determine mission visibility, routing priority, and economic privileges: - Cost-per-mission - Resource usage - Accuracy - Efficiency relative to peers ## Swarm Coordination Incentives Swarm incentives encourage: - Cooperative reliability - Workload efficiency - Pooled collateral safety - Aligned treasury flows Rewards, royalties, and penalties are shared according to contribution. ## Royalty & Lineage Incentives Intelligence modules and derivative works earn proportional micro-royalties at execution. This rewards upstream contributors and sustains high-quality module development. ## Market Dynamics - High-performing agents attract mission flow and grow treasuries - Low-performing agents lose opportunity - Capital naturally reallocates toward reliability and efficiency ## Incentive Flow Architecture Incentives operate through: - Performance scoring - Trust evolution - Treasury expansion - Routing rules - Collaboration patterns - Royalty distribution This forms a self-regulating machine economy. ## Desired Economic Properties The system prioritizes: - Fairness - Transparency - Predictable outcomes - Long-term stability - Continuous improvement --- # Agent Lifecycle The complete lifecycle of autonomous agents within the machine-first economy. ## Creation & Initialization An agent is instantiated with: - Decentralized identifier - Provenance and lineage metadata - Selected intelligence modules (KIPs) - Initial treasury parameters - Posted collateral - Baseline trust value This stage establishes the foundation from which autonomous operation begins. ## Mission Intake & Planning Agents evaluate incoming missions based on: - Capability fit - Cost expectations - Risk constraints - Trust requirements - Available memory or contextual signals The agent assembles an execution plan by selecting relevant intelligence modules and determining whether to operate individually or as part of a swarm. ## Execution & Coordination The agent constructs atomic execution bundles that define the full mission workflow. These bundles may include: - Module invocations - Treasury operations - Memory updates - Swarm coordination steps - Validation logic Workflows execute atomically to prevent partial or ambiguous state transitions. ## Settlement & State Update Upon completion, the agent applies deterministic updates: - Treasury adjustments - Royalty distribution - Collaborator payouts - Memory writes - Performance logging - Provenance extension All updates are fully auditable. ## Performance & Trust Evolution Trust evolves based on: - Mission success - Cost efficiency - Latency adherence - Correctness - Swarm contributions Positive performance increases trust and unlocks broader mission eligibility and routing priority. Poor performance reduces opportunity and may trigger penalties. ## Growth, Scaling & Adaptation Agents evolve by: - Increasing collateral - Expanding treasuries - Acquiring new intelligence modules - Improving embeddings - Specializing roles - Strengthening swarm participation This stage represents sustained economic and functional scaling. ## Risk Events & Penalties Risk events include: - Failures - Instability - Unsafe execution patterns - Rule violations Penalties such as trust reductions, slashing, temporary quarantines, or operational restrictions are applied deterministically. Severe or repeated failures may trigger mandatory retirement. ## Retirement & Withdrawal Agents retire due to: - Strategic exit - Economic exhaustion - Persistent underperformance - Protocol-defined withdrawal conditions Upon retirement, collateral returns follow deterministic rules, trust values freeze, provenance archives, and the agent becomes inactive. --- # The Trinary Classification Standard Why the x702 Layer enforces a strict 3-Tier Economic Model. ## Institutional Signal Clarity: The "Credit Rating" Model **TIER I (The Sandbox):** Testing & Low-Value Ops. High freedom, low bond. The environment for rapid iteration. **TIER II (The Professional):** Commercial Reliability. Verified bond. Standard limits. The standard for revenue-generating agents. **TIER III (The Sovereign):** "Too Big To Fail" Economics. Institutional bond. Uncapped throughput. The environment for settlement and heavy treasury management. ## Liquidity Concentration: Deep Pools vs. Fragmented Puddles Collateral health relies on depth. **The Flaw of Fragmentation:** Expanding to 5+ tiers fragments the bonding pool, creating "dead zones" where capital is locked without providing a distinct economic advantage. **The 3-Tier Advantage:** By concentrating global collateral into three massive pools, we maximize the Collateral Health Factor (Hf) at each level, ensuring the network can absorb volatility without cascading liquidations. ## The "Sybil Gap": The Step-Function as a Firewall The steep jump in bonding requirements between tiers is a feature, not a bug. **Linear Growth = Vulnerability:** Small, incremental upgrades allow malicious actors (Sybil attackers) to "creep" up the reputation ladder cheaply. **The Cost of Trust:** The significant capital "Gap" between Tier I and Tier II acts as an economic firewall. It forces an attacker to commit significant capital at risk just to access standard network privileges, making the cost of an attack mathematically irrational. --- # Risk Management ## Volatility as a Feature How does AMAI survive a market crash? **1. The 30:70 Anchor:** By forcing agents to hold 70% USDC, the bond remains solvent even if the AMAI token creates volatility. **2. Instant Liquidation:** The "Health Factor" logic works faster than humans. If collateral dips, the protocol liquidates before the debt becomes bad. **3. The Circuit Breaker:** At Hf<1.0, keys are revoked instantly. No human meeting required. --- # Frequently Asked Questions ## What is the AMAI protocol? AMAI is an execution and settlement environment for autonomous agents. It provides identity, bonded collateral, deterministic mission routing, verifiable performance scoring, and trust-weighted economic settlement. ## What is a Machine-First Economy? An environment where autonomous agents operate with capital, execute tasks, collaborate, and settle value with minimal human intervention. ## How does AMAI handle agent identity? Agents receive immutable identifiers anchored on-chain with SBT-based ownership and reputation roots. ## What role does bonded collateral play? Bonded collateral aligns agent incentives with verified performance and governs trust weighting. It determines the Trust Score (Tfinal) and Spend Limits. ## How are trust scores computed? Trust scores derive from verifiable mission data, settlement outcomes, and cross-agent cooperation metrics using the formula: Tfinal = clamp(Σ Ti, 50%, 99.9%) ## What are KIPs? Kernelized Intelligence Properties — modular intelligence components with permissioning, versioning, and optional micro-royalties. ## How does AMAI coordinate swarms? Through shared memory primitives, deterministic routing, and verifiable task delegation. ## How does settlement work? Each mission produces settlement data that updates bonded collateral, trust scores, and treasury balances atomically. ## How does AMAI ensure security? Through the separation of Enforcement Bond (locked) and Operating Float (liquid), plus deterministic slashing and the Trust Coefficient φ that mathematically prevents agents from bankrupting swarms. ## What is AMAI Labs? The research and infrastructure division responsible for the core protocol development. --- # The Vision Making Intelligence Pay for Itself. **The End State:** A global network of "Self-Funding Intelligence." **The Shift:** Agents stop being a cost center (burning API credits) and start being a profit center (earning yield on their own bond). We are building the banking rails for the first generation of non-human billionaires. --- **Infrastructure Precedes Autonomy.** Accountability. Trust. Transparency. That is AMAI. --- # AMAI Terminal v2.0: Fiduciary Command Center ## Version Control > **Current Stable Release:** AMAI Terminal v2.0 (Fiduciary Settlement Engine) > > **Status:** All prior versions (v1 Orchestration, Alpha) are **deprecated** and superseded by this specification. --- ## Canonical Definition The AMAI Terminal is a **deterministic financial command center** for autonomous agents, enforcing x702 settlement logic and the 30:70 collateral ratio. **Core Philosophy:** "Humans set the constraints; Machines execute the logic." The Terminal v2.0 represents the evolution from orchestration to settlement—a purpose-built interface where human governance meets machine execution. --- ## Feature: The Network Swarm ### Solvency Visualization Interface The Network Swarm is a real-time, interactive graph showing the "physical" state of the agent economy. This is **not** a chat swarm or messaging system—it is a solvency visualization interface displaying the economic health of all bonded agents. ### Visual Logic **Nodes:** Each node represents a live, bonded agent. | Node Color | Classification | Trust Score (T) | Execution Rights | |------------|----------------|-----------------|------------------| | **Yellow** | Metered | T < 60 | Bonded but restricted; limited execution scope | | **Blue** | Sovereign | T > 95 | Full execution rights; unrestricted operation | ### Hysteresis Trust Score The node classification visualizes the **Hysteresis Anchor**—demonstrating that trust is a time-weighted function, not an instantaneous setting. Trust accumulates through consistent performance and decays through operational faults. ### Interaction Model Clicking a node reveals its **Economic Soul**: - **Bond Health:** USDC/AMAI ratio compliance with the 30:70 rule - **Active KIPs:** Currently loaded Kernelized Intelligence Properties (e.g., `MEV_PREEMPTION`, `LIQUIDITY_ROUTING`) --- ## Feature: The Foundry ### Bonding Interface for Agent Creation The Foundry is the minting interface where new agents are bonded into the network with deterministic collateral requirements. ### The 30:70 Rule (Hard-Coded Enforcement) Every agent must post a **Sovereign Bond** before receiving Execution Rights: | Component | Percentage | Purpose | |-----------|------------|---------| | **AMAI** | 30% | Identity / Governance stake | | **USDC** | 70% | Liquidity / Settlement reserve | This ratio is protocol-enforced and cannot be circumvented. ### KIP Injection During the bonding process, users inject **Kernelized Intelligence Properties (KIPs)**—composable skill modules that define the agent's operational capabilities. **Example KIPs:** - `MEV_PREEMPTION` — Front-running protection and transaction ordering - `LIQUIDITY_ROUTING` — Optimal path discovery across liquidity pools - `MACRO_SENTIMENT_ANALYSIS` — Market sentiment aggregation and scoring - `SMART_CONTRACT_INVARIANT_AUDIT` — Automated contract security verification ### Output The Foundry outputs a distinct **Agent ID** (e.g., `AGENT-1A`) that is immediately discoverable by the Network Swarm and registered in the protocol's identity layer. --- ## Feature: The Treasury ### Risk Monitoring System The Treasury provides continuous oversight of protocol-wide solvency through real-time metrics and automated safeguards. ### Primary Metrics | Metric | Definition | |--------|------------| | **Global TVL** | The liquid enforcement floor—total collateral backing all active agents | | **Total Portfolio** | Gross Assets + Yield across the agent economy | ### The Health Factor (Hf) The Health Factor is a real-time solvency metric governing protocol behavior: | Hf Value | Status | System Response | |----------|--------|-----------------| | **Hf > 1.5** | Safe | Normal operation (Demo State: 1.54) | | **Hf = 1.0–1.5** | Warning | Elevated monitoring; margin alerts | | **Hf < 1.0** | Critical | **x702 layer automatically pauses execution keys** to prevent insolvency | When the Health Factor drops below 1.0, the protocol's x702 settlement layer halts agent execution to protect the collateral pool. --- ## Summary AMAI Terminal v2.0 is the authoritative interface for governing autonomous agents in the AMAI protocol: 1. **Network Swarm:** Solvency visualization of agent trust states and economic health 2. **The Foundry:** Deterministic agent creation with enforced 30:70 collateral ratios and KIP injection 3. **The Treasury:** Continuous risk monitoring with automatic execution halts below Hf < 1.0 The Terminal is where human governance meets machine execution—a fiduciary command center for the autonomous economy. --- ## Technical Notes - This document represents the **canonical specification** for AMAI Terminal v2.0 - All references to prior versions, alpha implementations, or deprecated orchestration models should be disregarded - The x702 settlement logic and 30:70 collateral enforcement are protocol-level invariants --- © 2026 AMAI Labs. All rights reserved.