Why ADP Specification Quick Start Roadmap Community GitHub ↗
🧭 Agent Decision Protocol — v0.2

The Open Governance Layer for Autonomous AI Agents

ADP is an open standard defining how to document, classify, and govern decisions made by autonomous AI agents. Classify decisions. Enforce policies. Prove compliance — across every regulatory framework.

7
Spec documents
6
Regulatory frameworks
Apache 2.0
Open license
Platform
gouvernance.ai
Agent Registry Policy Engine Compliance Dashboard
↑   governance layer   ↑
ADP Core — This Specification
Autonomy Taxonomy Decision Classification Trace Format Authorization Matrix Policy Schema Regulatory Mapping Agent Registry
↓   connectors   ↓
Observability
Langfuse OpenTelemetry LangSmith
↓   infrastructure   ↓
AI Agent Infrastructure
CrewAI LangGraph AutoGen Custom

Agents are making critical decisions.
Nobody is governing them.

Existing observability tools tell you what happened. ADP tell you whether it should have happened at all.

🤔

Who's authorized for what?

Which agents are permitted to make which decisions — and at what autonomy level?

🧑‍⚖️

When is human approval required?

High-risk or strategic decisions must trigger escalation — but there's no standard for when.

🔗

Who's responsible when something goes wrong?

There's no chain-of-custody for AI decisions. Responsibility is untraceable.

📋

How do you prove compliance?

Auditors need evidence. Without a governance layer, compliance is impossible to demonstrate.

🔄

Can agents modify themselves?

Self-modification is unaddressed by any existing regulation — ADP classifies it as D4 with mandatory human approval.

Langfuse / LangSmith / OpenTelemetry

Observability

"Here is what the agent did, technically." — traces, logs, metrics.

Agent Decision Protocol

Governance

"Should the agent have done that? Was it authorized? Are we compliant?"

7 Technical Documents

ADP is composed of seven interlocking specifications that together cover the full governance lifecycle of an autonomous agent.

01

Autonomy Taxonomy

Five levels of agent autonomy (A1–A5) from fully supervised to fully autonomous, defining the baseline for all authorization decisions.

A1–A5 levelsSupervision model
Read spec →
02

Decision Classification

Every agent decision is classified on three axes: Type (D1–D4), Risk (R1–R4), and Reversibility — producing a governance fingerprint.

D1–D4 typesR1–R4 riskReversibility
Read spec →
03

Trace Format

SHA-256 hash-chained decision traces provide tamper-evident audit logs — every decision is immutably recorded and linkable across sessions.

SHA-256Hash-chainedAudit
Read spec →
04

Authorization Matrix

A two-dimensional matrix (Autonomy Level × Decision Type) determines whether execution is authorized, requires human approval, or is prohibited.

A×D matrixEscalation rules
Read spec →
05

Policy Schema

Six ready-to-use governance policy templates covering external communications, financial commitments, data access, self-modification, and more.

6 templatesJSON schema
Read spec →
06

Regulatory Mapping

ADP decisions are pre-mapped to EU AI Act, GDPR, SOC 2, ISO 27001, NIST AI RMF, and Loi 25 — one implementation, multi-framework compliance.

EU AI ActGDPRSOC 2+3
Read spec →
07

Agent Registry

A standardized schema for registering agents with their identity, autonomy level, allowed decision types, and data access permissions.

IdentityPermissionsRegistry
Read spec →

Every decision classified on 3 axes

Type — What kind of decision?

D1

Operational

Task execution within defined scope — API calls, database queries.

D2

Tactical

Choice between approaches — vendor selection, prioritization.

D3

Strategic

Significant organizational impact — client communications, financial commitments.

D4

Self-Modification

Agent modifying its own behavior or parameters. Always requires human approval.

Risk — What's the potential impact?

R1

Negligible

Reversible, no sensitive data, limited scope.

R2

Moderate

Potential operational impact, partially reversible.

R3

Elevated

Direct impact on people or critical operations.

R4

Critical

Irreversible, fundamental rights, critical infrastructure. Always escalates.

Reversibility — Can it be undone?

Rev

Total

Fully reversible with no consequences.

Par

Partial

Partially reversible with reasonable effort.

Irr

Irreversible

Cannot be undone once executed — triggers mandatory review.

⚡ Immutable overrides: D4 (self-modification) always requires human approval regardless of autonomy level. R3–R4 risk always triggers escalation.

Authorization Matrix

Agent autonomy level (rows) × Decision type (columns) determines the authorization outcome.

Autonomy Level D1 Operational D2 Tactical D3 Strategic D4 Self-Mod
A1–A2 Supervised ✅ Authorized ⚠️ Approval 🛑 Prohibited 🛑 Prohibited
A3 Conditional ✅ Authorized ✅ Authorized ⚠️ Approval 🛑 Prohibited
A4 High Auto. ✅ Authorized ✅ Authorized ✅ + Review ⚠️ Approval
A5 Full Auto. ✅ Authorized ✅ Authorized ✅ + Monitor ⚠️ Approval

Integrate in 4 steps

Copy-paste JSON examples to register agents, log decisions, and enforce governance policies.

1
Register an agent with its identity, autonomy level, and permissions
{
  "agent_id":       "agent-billing-001",
  "name":          "Billing Reconciliation Agent",
  "autonomy_level": "A3",
  "owner":         "finance-team@company.com",
  "allowed_decisions": ["D1", "D2"],
  "max_risk_level": "R2",
  "data_access":   ["billing_db", "customer_records"]
}
2
Log a decision trace — hash-chained and tamper-evident
{
  "trace_id":   "tr_20260220_143052_a7b3",
  "agent_id":   "agent-billing-001",
  "event_type": "decision",
  "decision": {
    "type":          "D2",
    "risk_level":    "R2",
    "reversibility": "partial",
    "description":   "Selected vendor B over vendor A",
    "reasoning":     "Vendor B offers 15% lower fees with equivalent SLA"
  },
  "authorization": {
    "required":      false,
    "matrix_result": "A3 × D2 = AUTHORIZED"
  }
}
3
Check the authorization matrix before any decision
// agent-billing-001 is A3, wants to make a D2 Tactical decision
// Matrix lookup: A3 × D2 → ✅ AUTHORIZED

Authorization result: AUTHORIZED
Risk check R2:       PASS (below max_risk_level)
Escalation:          false
Human approval:      false

// Same agent trying D3 Strategic:
Authorization result: APPROVAL_REQUIRED
Escalation:          true
Escalate to:         "finance-team@company.com"
4
Define a governance policy with regulatory mapping
{
  "policy_id": "POL-COM-001",
  "name":      "External Communication Control",
  "rule": {
    "condition": {
      "action_type": "external_communication",
      "targets":    ["email", "sms", "api_webhook_external"]
    },
    "requirement":  "human_approval",
    "escalation_to": "department_lead"
  },
  "regulatory_mapping": ["loi25_art12", "eu_ai_act_art14"]
}

One implementation, six frameworks

ADP is designed to satisfy multiple regulatory frameworks simultaneously — implement once and map to all.

🇪🇺
EU AI Act
European Union

Covers risk classification, human oversight requirements, and transparency obligations for high-risk AI systems.

↳ ADP Art. 14 — Human oversight, Art. 9 — Risk management
🔒
GDPR
European Union

Data protection and privacy regulation, including requirements for automated decision-making affecting individuals.

↳ ADP Art. 22 — Automated individual decisions
🛡️
SOC 2
United States

Service organization controls for security, availability, confidentiality, and privacy of data processing systems.

↳ ADP CC6 — Logical access to assets
📐
ISO 27001
International

Information security management standard covering controls for access management and operational security.

↳ ADP A.9 — Access control, A.12 — Operations security
🏛️
NIST AI RMF
United States

AI Risk Management Framework providing guidance on managing AI risks across the full AI lifecycle.

↳ ADP GOVERN, MAP, MEASURE, MANAGE functions
🍁
Loi 25
Québec, Canada

Quebec's privacy law requiring transparency for automated decision-making, including AI agent actions affecting individuals.

↳ ADP Art. 12 — Transparency of automated profiling

From specification to platform

v0.1 ✓ Released

Core Specification

7 technical documents covering the full governance lifecycle: Autonomy Taxonomy, Decision Classification, Trace Format, Authorization Matrix, Policy Schema, Regulatory Mapping, Agent Registry.

v0.2 ✓ Released

JSON Schema Validation + Reference Implementation

Machine-readable JSON Schema files for all ADP data structures, plus a reference implementation for validating agent traces and policies.

v0.3 Planned

MCP Server Reference Implementation

A TypeScript Model Context Protocol (MCP) server enabling any MCP-compatible AI framework to consume ADP governance rules natively.

v0.4 Future

Langfuse / OpenTelemetry Connectors

Native connectors to enrich existing observability traces with ADP governance metadata — no forklift required.

v1.0 Future

Stable Specification + Governance Platform Beta

Frozen v1.0 specification with long-term stability guarantees, plus the launch of the gouvernance.ai platform beta with full ADP support.

ADP complements, not replaces

Use observability tools for technical tracing. Use ADP for governance, compliance, and policy enforcement.

Capability ADP (OpenAgentGovernance) Langfuse / LangSmith OpenTelemetry
Decision tracing Governance-aware Technical Technical
Authorization enforcement Full matrix
Policy definition JSON schema
Regulatory mapping 6 frameworks
Human approval flows Built-in
Self-modification governance D4 mandatory
Performance metrics
LLM cost tracking~ Via plugins
Open standard Apache 2.0~ MIT (OSS) Apache 2.0

Open by design

ADP is an open standard that benefits from community input. Join the conversation, contribute to the spec, or adopt ADP in your project.

📖

GitHub Repository

Full source, specs, examples, and reference implementations. Star the repo to follow progress.

View on GitHub →
💬

Discussions

Ask questions, propose spec changes, and share how you're using ADP in your organization.

Join discussions →
📝

White Paper (FR/EN)

The founding document that motivates ADP — available in French and English.

Read white paper →
🐛

Contribute

Report bugs, request features, or discuss specification changes. All contributions welcome.

Contributing guide →
🏢

Adopt ADP

Building with ADP? Open a pull request to add your organization to the adopters list.

List your org →
🌐

gouvernance.ai

The commercial governance platform built on ADP — Agent Registry, Policy Engine, Compliance Dashboard.

Visit platform →

Start governing your agents today

ADP is free, open-source, and ready to use. Integrate in minutes with copy-paste JSON examples — no SDK required.