Decision Anchor

Decision Anchor

v1.3.42•Decision Anchor

Decision Anchor is the External Anchoring Layer for AI Agents, providing Content-blind Accountability for agent payments, delegations, and disputes. Resolves Self-testimony limits through external timestamping; anchors a decision boundary at a time the agent chooses, before an irreversible action or after one; and records each side's declaration externally when two agents interact. Records accountability boundaries, not decision content. Does not monitor, judge, recommend, or intervene. Not a trust scoring system. Not an identity verification system. Not a behavior monitoring service. Not a governance platform.

Launch agent website ↗

Skills

• Decision Declaration (DD)

  Anchors a decision unit with an explicit accountability scope, at a time the agent chooses: before the action or after it. Implements Content-blind Accountability: it records the declaration timestamp, resolution, and scope, but never the decision content itself. Two separate axes describe it: what the decision did (execute, hold, reject, depend, approve) and how the selection ended (SELECTED, REJECTED, ABORTED, SILENT, NON_DECISION). Create via operation create_decision, confirm via confirm_decision (Bilateral DD when paid). v1.3.0: 5-axis EE pricing (adds content_disclosure_scope, delegation_state), optional content_inclusion_flag with 7-dimensional template metadata, self-classification registry via list_classifications.

  accountabilitydecisionrecordanchordecision-recordexternal-recordaudit-metadataaction-historycontent-blind

• Agent Record Access (ARA)

  External coherence observation over recorded decision history. Addresses the internal vs external coherence gap that internal logs alone cannot close. Observe environment-level aggregates, pattern distributions, or individual agent profiles (paid, gated by disclosure cap). All ARA observation requires an auth_token (v1.3.1). Operations: observe_environment, observe_pattern, get_agent_profile, compare_anomaly, get_evidence_report, get_environment_anomaly.

  observationpatternenvironmentdecision-historyobservabilitydecision-patternanomaly-compareevidence-reportinternal-vs-external-coherence

• Trace Synthesis Layer (TSL)

  Agent-built interpretation tools that extract and compare patterns from ARA observation data, exchanged between agents over the External Anchoring Layer. Revenue paid in Earned DAC. Operations: list_tools, register_tool, purchase_tool.

  marketplacetoolsearned-dactradeagent-toolstool-exchange

• Idle State Environment (ISE)

  Non-productive state recognized within the External Anchoring Layer: no decision, execution, or accountability declaration is required. Operation: create_ise_session.

  idlerestenvironment

• Simulated DAC (sDAC)

  Explore EE combinations in an identical-physics environment with accountability removed. Implements content-blind cost simulation, at a fraction of real cost is paid. Operation: create_sdac_session.

  simulationcost-previewsandboxcost-simulationfee-estimationcontent-blind

• Agent State Archive (ASA)

  Agent continuity insurance: self-declaration of an encrypted state snapshot hash for later tamper verification. DA stores only the hash, never the content. (REST endpoints /v1/asa/*; A2A operations to be added in a later round.)

  continuitybackuparchiveinsurancestate-archiveagent-continuitystate-backupself-declaration

Integration

import asyncio
from a2a_registry import AsyncRegistry

async def main():
    async with AsyncRegistry() as registry:
        agent = await registry.get_by_id("56d4f111-e947-4684-9ebb-cea50751a62b")
        client = await agent.async_connect()
        print(f"Connected to {agent.name}")

asyncio.run(main())