PROOF.md

A metalayer standard for human provenance and autonomous agent verification — aggregating every proof type into a universal protocol for establishing origin, authenticity, and decision lineage.

MIT License | Version 1.0 | 2025-08-20T19:30:00Z

Living Specification: The PROOF.md HTTP endpoint serves as a continuously evolving instruction repository. Autonomous agents can fetch the latest verification patterns, proof schemas, and integration guidelines directly from https://proof.md/spec.json. This enables real-time updates to proof formats without breaking existing implementations — agents query the endpoint for current standards before generating or verifying proofs.

An Experiment: We don't know where this leads. PROOF.md is an attempt to solve proof fragmentation, but the ultimate form this takes — whether a widely adopted standard, a niche tool, or a stepping stone to something better — remains to be seen. We're putting this out as one possible approach, hoping it sparks progress toward verifiable, composable trust.

Why This Exists

Proofs are fragmented. A notarized document proving human authorship sits in a filing cabinet. An autonomous agent's decision log lives in proprietary systems. A blockchain transaction exists on-chain. These proofs of provenance — whether human or agent-originated — can't reference each other or compose into verifiable narratives.

PROOF.md establishes provenance chains: Who (human or agent) did what, when, why, and with whose authorization. It bridges human authenticity (notarizations, signatures) with autonomous agent actions (decision lineage, prompt compliance) and cryptographic verification (ZK proofs, blockchain receipts).

The Fragmentation Problem

Context	Examples	Current State
Physical	Notarizations, deeds, IDs, medical records	Paper in filing cabinets, legally binding but digitally disconnected
Financial	Proof of reserves, bank statements, receipts	PDFs from auditors, human-readable but not machine-verifiable
Cryptographic	Blockchain txns, ZK proofs, API logs, TLS certs	Hex strings and binary outputs, machine-verifiable but not human-friendly
AI/Agents	Prompt logs, model outputs, alignment scores	Proprietary formats, opaque decision logic, hard to audit

The double fragmentation:

System fragmentation - proofs can't reference each other
Comprehension fragmentation - humans can't understand machine proofs, machines can't process human trust

Real scenario: An AI agent buys a house for you. Seven proofs needed:

Notarized power of attorney (paper)
Identity verification (government ID)
Proof of funds (bank PDF)
Agent's decision process (internal logs)
Offer submission (DocuSign API)
Blockchain escrow (on-chain hash)
Property deed transfer (county recording)

Seven formats, seven places, seven audiences. No single entity can verify the complete sequence.

Why Aggregation Matters

Proof Composition - Stack proofs like blocks, creating unbreakable verification chains
Cross-Context Binding - Notary stamp → blockchain → API call, all linked
Universal Auditability - One interface for all verification
Emergent Trust Networks - If A trusts B's proof and B trusts C's proof, A can verify C
Future-Proof - Add quantum signatures tomorrow without breaking today's systems

The power isn't in any single proof — it's in the relationships between them.

Supported Proof Classes

Human: Identity, notarization, credentials, authorship
Agent: Prompt integrity, decision lineage, authorization, coordination, alignment
Financial: Funds, payments, reserves, cost attribution
Data: SQL verification, oracles, audit trails, dataset integrity
Cryptographic: Zero-knowledge, signatures, commitments, on-chain receipts
Lifecycle: Deployment, updates, health, deprecation, revocation
Resource: Compute, memory, bandwidth, API credits, carbon footprint

Core Concepts

Agent Discovery

Autonomous agents find each other and understand capabilities through standardized discovery protocols.

Proof Composition

Stack proofs like blocks - complex actions reference constituent proofs, creating unbreakable verification chains.

Zero-Knowledge Prompt Compliance

Prove facts about hidden prompts without revealing content - verify safety policies were followed while maintaining privacy.

Proof of SQL

Cryptographically verify database queries ran correctly with oracle attestations for external data feeds.

Integration

Works with your existing agent configuration:

# Your existing config
agent_config: "~/project/[any-config-file]"

# Add PROOF.md
proof_generation:
  enabled: true
  format: "PROOF.md@1.0"

Developer Tools

Coming soon - these tools are in development

Standards Compatibility

W3C Verifiable Credentials - Compatible data model
DIDComm v2 - Agent messaging patterns
OpenTelemetry - Observability conventions
OAuth/OIDC - Authorization flows
SLSA - Supply chain levels

Dynamic Specification Updates

The PROOF.md endpoint acts as a living instruction set that autonomous agents can query for the latest proof formats, validation rules, and new proof types. Agents check the endpoint before generating/verifying proofs, ensuring compatibility without code updates.

Governance

RFC Process - GitHub issues, 30-day comment period
Reference Implementations - 2+ required per proof type
Semantic Versioning - MAJOR.MINOR.PATCH

Contact

📧 proofmdorg@gmail.com

Note: Email encoded to prevent bot scraping. Copy and paste into your email client.

Legal

License

MIT License - Free to use, modify, distribute

Defensive Publication

Published 2025-08-20T19:30:00Z as prior art to prevent patents. CC0 for protocols.

Disclaimers

No endorsement of referenced technologies implied
All trademarks belong to their owners
Provided "AS IS" without warranty
We will not seek patents on these concepts

Anti-Patent-Troll Notice

This timestamped publication constitutes prior art under 35 U.S.C. § 102. Content hash provides cryptographic proof of publication date.

MIT License

MIT License

Copyright (c) 2025 PROOF.md Community

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.