WebMEM™

The Protocol for Structuring, Delivering, and Conditioning Trust-Scored AI Memory on the Open Web

RFC-002 — WebMEM Document Specification (v0.1)

A Protocol for Machine-Retrievable, Trust-Anchored Knowledge Units

Metadata

  • rfc_id: RFC-002
  • title: WebMEM Digest Document Specification
  • status: Draft
  • version: 0.1
  • authors:
    • David W. Bynon (@TrustPublishing)
    • WebMEM Working Group
  • date_created: 2025-07-15
  • license: CC BY-SA 4.0
  • domain_scope: General (Vertical-Agnostic)
  • depends_on: RFC-001

Purpose

Defines the formal structure of a WebMEM Digest—a modular, trust-scored memory object containing machine-ingestible fragments, glossary references, and multi-format export paths. WebMEM Digests support AI retrievability, explainability, and citation integrity across platforms and agents.

Core Structure: SemanticDigest

semantic_digest: cms-ma-mapd-plan
fragment_scope: plan
html_url: https://medicarewire.com/az/2025/H1234/001/
citation_policy:
  cite: html_url
  ingest: yaml
provenance_scope: multi-dataset

plan_summary:
  contract_id: H1234
  plan_id: 001
  segment_id: 0
  cms_plan_identifier: H1234-001-0
  plan_name: "AwesomeCare Medicare Plus"
  coverage_year: 2025
  fips_county_code: 04013
  county_name: "Maricopa"
  state: AZ
  star_rating: 4.5

monthly_costs:
  premium:
    data_id: in_premium
    value: "$0"
    unit: usd
    derived: false
  moop:
    data_id: moop
    value: "$5,500"
    unit: usd
    derived: true
    derived_from: pbp_b7_health_prof

provenance:
  - dataset_id: CMS_PBP_2025
    dataset_title: CMS Plan Benefit Package 2025
    source_agency: CMS
    license: Public Domain
    published: 2025-07-01
    retrieved: 2025-07-06
    trust_layer: primary
    dataset_home: https://cms.gov/pbp

structured_data_formats:
  jsonld: https://medicarewire.com/digests/H1234-001-0/digest.jsonld
  ttl: https://medicarewire.com/digests/H1234-001-0/digest.ttl
  prov: https://medicarewire.com/digests/H1234-001-0/digest.prov
  xml: https://medicarewire.com/digests/H1234-001-0/digest.xml
  md: https://medicarewire.com/digests/H1234-001-0/digest.md

Required Fields

Field Description
semantic_digest Unique ID for this digest
fragment_scope Scope of entity (e.g., plan, provider, term)
html_url Canonical human-facing location
citation_policy Machine citation + ingestion instructions
provenance List of datasets and metadata (see RFC-003)
structured_data_formats Map of export formats for memory ingestion
content_block(s) At least one (e.g., plan_summary, glossary, etc.)
mapped data_ids All IDs must conform to RFC-001 definitions

Design Principles

  • Machine retrievability
  • Explainability + auditability
  • Dataset anchoring
  • Fragment-level scoping
  • AI-ingestible across formats

Format Use Cases

Format Use Case
.yaml Human-readable canonical digest
.jsonld AI ingestion + schema-based validation
.ttl RDF / knowledge graph integration
.prov Trust lineage + machine-verifiable provenance
.xml Legacy system interoperability
.md Developer-facing Markdown documentation
.owl Inferential logic + AI rule encoding

Example Directory Layout

/semantic-digests/plan/H1234-001-0/
├── digest.yaml
├── digest.jsonld
├── digest.ttl
├── digest.prov
├── digest.xml
└── digest.md

Digest-Level Trust Contract

A WebMEM Digest is trust-qualified only if:

  • All data_ids conform to RFC-001
  • Derived values declare source fields and logic
  • At least one provenance block is included
  • At least one export format is memory-ingestible

Canonical Reference

This RFC is permanently located at webmem.com/rfc/rfc-002/ and is versioned in the WebMEM GitHub Registry.