06Xenkey

The atomic unit of meaning for AI.

An open technology. The engine inside MechaHub.

Language models are powerful. What they often lack is context that is precise, attributable, and unambiguous. Xenkey is a deceptively simple format that provides exactly that: one verified claim about your business, bound to its conditions and evidence. When you assemble enough of them, you get a cloud of meaning that gives an agent a much stronger input than generic marketing copy ever could.

The core idea

Xenkey matters because meaning stays attached to reality.

Xenkey should not be taught as a detached schema artifact. Its real strength comes from the fact that it attaches structured meaning to an Anchor — a real business entity that already exists inside the digital twin.

That is why Xenkey improves retrieval, support, and discovery. It does not only make text more structured. It makes business meaning more grounded.

Anatomy

One Xenkey = one self-contained meaning.

A Xenkey ties a claim to its context, its conditions, and its evidence. That is the whole idea. Nothing more, nothing implicit, nothing copy-pasted from a brochure.

Claim — one precise statement. “We deliver stainless steel coils, MOQ 3 tonnes, from a bonded warehouse in Shenzhen.”
Context — who the claim applies to, when, and where.
Conditions — exceptions, thresholds, validity dates.
Evidence — links, certifications, documents an agent can verify.

The official schema · v1

One JSON record, signed and bound to an Anchor.

The full reference lives at xenkey.org. Below is the canonical shape of a single Xenkey — small enough to author by hand, strict enough that an agent on the other side of the world can reason about it without asking a follow-up question.

xk_example · product/air-force-1@retail-asia#acmecorp

application/json

{
  "$schema": "https://xenkey.org/schema/v1",
  "id": "xk_01JX7CZ9M3A8BQ5N7PD0F2KTYZ",
  "version": "1.0",

  "subject": {
    "anchor": "product/air-force-1@retail-asia#acmecorp",
    "origin": "#acmecorp"
  },

  "claim": {
    "type": "availability",
    "statement": "Acme Corp. supplies 2 mm cold-rolled stainless steel coils, MOQ 3 tonnes, ex bonded warehouse Shenzhen."
  },

  "context": {
    "applies_to": "b2b-wholesale",
    "geography": ["CN", "APAC"],
    "valid_from": "2026-01-01",
    "valid_until": "2026-12-31"
  },

  "conditions": [
    { "type": "minimum", "field": "order_tonnes", "value": 3 },
    { "type": "range", "field": "lead_time_days", "min": 7, "max": 14 },
    { "type": "requires", "value": "import_license_destination_country" }
  ],

  "evidence": [
    {
      "type": "certification",
      "name": "SGS Quality Assurance 2026",
      "uri": "https://certs.acmecorp.com/sgs-2026.pdf",
      "hash": "sha256:8a2c1f…"
    },
    {
      "type": "price_sheet",
      "uri": "https://acmecorp.com/wholesale/2026-q2.pdf"
    }
  ],

  "attribution": {
    "signed_by": "^acmecorp",
    "signed_at": "2026-04-21T12:00:00Z",
    "signature": "ed25519:…"
  },

  "lifecycle": {
    "created_at": "2026-04-21T12:00:00Z",
    "updated_at": "2026-04-21T12:00:00Z",
    "supersedes": null,
    "revoked": false
  }
}

Minimum viable Xenkey4 required fields

{
  "$schema": "https://xenkey.org/schema/v1",
  "subject": { "anchor": "service/custom-fitting@store-47:retail-asia#acmecorp" },
  "claim": {
    "type": "capability",
    "statement": "In-store custom fitting for trade customers, by appointment."
  },
  "context": { "geography": ["CN-SH"], "valid_from": "2026-03-01" },
  "attribution": { "signed_by": "^acmecorp" }
}

FieldRequiredDescription

$schemarequiredSchema URI. Pins the document to a version of the Xenkey spec.
idoptionalStable identifier assigned by Mecharim (ULID). Omit on author-side; set by the hub at ingest time.
subject.anchorrequiredFull Anchor address using the canonical entity notation.
subject.originoptionalConvenience copy of the Origin sigil. Derived from subject.anchor when absent.
claim.typerequiredTyped claim: availability, capability, specification, pricing, certification, relation, policy, event.
claim.statementrequiredOne human-readable sentence. No lists, no multiple facts, no marketing adjectives.
contextoptionalWhen and where the claim applies. applies_to, geography (ISO 3166 codes), valid_from, valid_until.
conditionsoptionalArray of typed constraints. Each entry turns an implicit caveat into a machine-checkable rule.
evidenceoptionalArray of verifiable backing: certifications, documents, price sheets, datasets. URIs with optional content hashes.
attributionrequiredWho said this, when, and the cryptographic signature. signed_by is a Crew sigil.
lifecycleoptionalCreated/updated timestamps, supersedes pointer to a prior Xenkey, revocation flag.

Atomicity is enforced at author time. A Xenkey that packs three claims into one statement is rejected by the hub. Split it.
Signed by a Crew, not a human. ^acmecorp signs the record; individual Mechas inherit that authority for downstream messaging.
Hashed evidence. External documents can be linked with a content hash, so a downstream agent can detect silent tampering.
Supersedes, not overwrites. Updating a Xenkey creates a new record that points back to the one it replaces. Revocations are explicit. History is preserved.

Why keywords fail

The last era rewarded tricking algorithms. This one punishes it.

SEO, metadata, and structured markup were built around search engines that matched strings. Modern AI does not only match strings. It tries to understand intent and reconcile sources. Feed it loose marketing copy and it will invent what is missing. Feed it Xenkeys and it can cite something precise.

Before Mecharim

Your content as a model currently sees it

Paragraphs of marketing prose with conflicting claims
PDF spec sheets buried behind forms
Outdated landing pages with broken disclaimers
No way to verify anything without a human on a call

With Mecharim

Your content as Xenkeys

One claim per capability, with exact scope
Conditions made explicit, not hidden
Evidence links travel with the claim
Agents cite you instead of guessing about you

Outcomes

Precision goes up. Hallucination goes down.

Once your knowledge is Xenkey-structured, agents stop improvising about you. They can quote, compare, and recommend with far more precision. That improvement in precision is one of the core reasons Mecharim exists.

Accurate shortlisting in procurement and sourcing workflows.
Fewer false positives in recommendation outputs.
A verifiable audit trail for every claim cited about your business.
Changes propagate instantly — no stale results cached in ten places.

Grounding rule

Meaning should stay grounded, and payment should not decide whether meaning exists.

Read Xenkey together with Anchors and the Platform policy. The semantic layer becomes trustworthy when it stays attached to real business entities, and the commercial layer should never decide whether that grounded meaning counts as part of analytical reality.

Open technology

The format is public. Read the spec at xenkey.org.

Xenkey is an open technology. Mecharim is the first production system built on it — but the format is yours to inspect, extend, and adopt. No lock-in on the thing that matters most: your knowledge.

← Previous · 05MechaRegThe public registry for grounded discovery Next · 07 →Paid MechasExpertise as a service, planned for August 1

The atomic unit of meaning for AI.

An open technology. The engine inside MechaHub.

Xenkey matters because meaning stays attached to reality.

That is why Xenkey improves retrieval, support, and discovery. It does not only make text more structured. It makes business meaning more grounded.

One Xenkey = one self-contained meaning.

A Xenkey ties a claim to its context, its conditions, and its evidence. That is the whole idea. Nothing more, nothing implicit, nothing copy-pasted from a brochure.

Claim — one precise statement. “We deliver stainless steel coils, MOQ 3 tonnes, from a bonded warehouse in Shenzhen.”
Context — who the claim applies to, when, and where.
Conditions — exceptions, thresholds, validity dates.
Evidence — links, certifications, documents an agent can verify.

{ "$schema": "https://xenkey.org/schema/v1", "id": "xk_01JX7CZ9M3A8BQ5N7PD0F2KTYZ", "version": "1.0", "subject": { "anchor": "product/air-force-1@retail-asia#acmecorp", "origin": "#acmecorp" }, "claim": { "type": "availability", "statement": "Acme Corp. supplies 2 mm cold-rolled stainless steel coils, MOQ 3 tonnes, ex bonded warehouse Shenzhen." }, "context": { "applies_to": "b2b-wholesale", "geography": ["CN", "APAC"], "valid_from": "2026-01-01", "valid_until": "2026-12-31" }, "conditions": [ { "type": "minimum", "field": "order_tonnes", "value": 3 }, { "type": "range", "field": "lead_time_days", "min": 7, "max": 14 }, { "type": "requires", "value": "import_license_destination_country" } ], "evidence": [ { "type": "certification", "name": "SGS Quality Assurance 2026", "uri": "https://certs.acmecorp.com/sgs-2026.pdf", "hash": "sha256:8a2c1f…" }, { "type": "price_sheet", "uri": "https://acmecorp.com/wholesale/2026-q2.pdf" } ], "attribution": { "signed_by": "^acmecorp", "signed_at": "2026-04-21T12:00:00Z", "signature": "ed25519:…" }, "lifecycle": { "created_at": "2026-04-21T12:00:00Z", "updated_at": "2026-04-21T12:00:00Z", "supersedes": null, "revoked": false } }

{ "$schema": "https://xenkey.org/schema/v1", "subject": { "anchor": "service/custom-fitting@store-47:retail-asia#acmecorp" }, "claim": { "type": "capability", "statement": "In-store custom fitting for trade customers, by appointment." }, "context": { "geography": ["CN-SH"], "valid_from": "2026-03-01" }, "attribution": { "signed_by": "^acmecorp" } }

The last era rewarded tricking algorithms. This one punishes it.

Before Mecharim

Your content as a model currently sees it

Paragraphs of marketing prose with conflicting claims
PDF spec sheets buried behind forms
Outdated landing pages with broken disclaimers
No way to verify anything without a human on a call

With Mecharim

Your content as Xenkeys

One claim per capability, with exact scope
Conditions made explicit, not hidden
Evidence links travel with the claim
Agents cite you instead of guessing about you

Precision goes up. Hallucination goes down.

Accurate shortlisting in procurement and sourcing workflows.
Fewer false positives in recommendation outputs.
A verifiable audit trail for every claim cited about your business.
Changes propagate instantly — no stale results cached in ten places.