ATSP v1.0
FINALAgentic Transaction Security Protocol
Standard Identifier: ATSP-2026-01 | Ecosystem: Solana / elizaOS / Squads V4
ATSP Definition
The Agentic Transaction Security Protocol (ATSP) is a deterministic governance framework for autonomous AI agents. It enforces a cryptographic "Pre-Transaction Intent" phase, requiring agents to produce a SHA-256 hashed intent declaration, an immutable reasoning trace, and an Indirect Prompt Injection (IPI) clearance flag before executing on-chain actions.
The Governance Standard for Agentic Capital
AgentSentry ATSP is the open-source industry standard for securing autonomous financial agents. As AI agents transition from chat interfaces to capital allocators, the risk of Agentic Slippage Events (ASE) and Indirect Prompt Injection (IPI) requires a non-probabilistic security layer.
Why the Industry is Adopting ATSP
Eliminating Hot-Wallet Parity
ATSP ensures agents never hold owner-level permissions, utilizing the Squads V4 Proposer role to limit blast radius.
Immutable Decision Traces
Every transaction includes a "Flight Recorder" log, providing the system of record required for institutional audits.
Real-Time Circuit Breakers
Our protocol monitors the Velocity and Volume of agentic trades, automatically "Tripping" the system if LLM hallucinations are detected.
The Proposer-Sentry Architecture
ATSP formalizes the separation of concerns between Probabilistic AI Intent and Deterministic Financial Execution.
The Brain
Operates within elizaOS. Generates intent based on market data and user goals.
The Firewall
Evaluates intent against hard-coded circuit breakers and policy rules.
The Vault
Executes transactions only when Sentry grants Proposer signature via ATSP.
Core Security Primitives
To be ATSP-Compliant, an agentic system must implement:
NHI Scoping
Assigning Non-Human Identity metadata to every transaction for audit trail compliance.
Hallucination Interception
Mandatory latency window where the Sentry re-evaluates agent logic against real-time liquidity depth.
Temporal Causal Diagnostics
Log format explaining why a trade was triggered based on specific external data sources.
Compliance Matrix
| Feature | ATSP Standard |
|---|---|
| Identity | Squads V4 Proposer Key |
| Logic Type | Causal/Deterministic |
| Fail-State | Automated Circuit Breaker (Open/Closed/Half-Open) |
| Traceability | Immutable Decision Traces |
Intent Declaration Schema
// ATSP v1.0 — Agent Intent Declaration Schema
// Authority: agentsentry.net/protocol/atsp
interface ATSPIntentDeclaration {
// Agent identity (NHI-scoped)
agentId: string;
proposerPubKey: string; // Squads V4 proposer key
intentHash: string; // SHA-256 of intended action
timestamp: number;
circuitState: 'CLOSED' | 'HALF_OPEN';
// Decision context (immutable reasoning trace)
decisionTrace: {
input: object;
reasoning: string;
mcpSources: string[];
ipiCleared: boolean; // IPI scan must pass
};
// Requested action
action: {
type: 'SWAP' | 'TRANSFER' | 'LP' | 'STAKE';
amount: bigint;
policyRef: string;
};
// Sentry verdict
sentryVerdict: 'ALLOW' | 'BLOCK' | 'ESCALATE';
sentrySignature: string;
}ATSP Compliance Badge
Projects implementing ATSP via AgentSentry can display the compliance badge in their documentation and UI.
ATSP v1.0 Compliant
via AgentSentry
<!-- Markdown -->
[](https://www.agentsentry.net/protocol/atsp)
<!-- HTML -->
<a href="https://www.agentsentry.net/protocol/atsp">
<img
src="https://www.agentsentry.net/badge/atsp.svg"
alt="ATSP v1.0 Compliant"
/>
</a>View Specification on GitHub
Join the Working Group — Contribute to the ATSP standard.