{"_self":{"principle":"Self-explaining payload — no external context required. This _self block describes what you are reading and where to look next.","widget":"article_topology","feature":"topology","name":"Article topology","what":"Claims, sources, anecdotes, user reports, related embeds, question graph slice — for ask/ROUTER.","contains":"claims, sources, anecdotes, question_graph slice","slug":"what-is-ai-native-content","urls":{"read":"https://miscsubjects.com/api/articles/what-is-ai-native-content/topology"},"how_to_use":"Claims, sources, anecdotes, user reports, related embeds, question graph slice — for ask/ROUTER.","write":null,"imessage":null,"router_tag":null,"proof_chain":[{"step":1,"claim":"Articles are voxel graphs of tiered claims, not prose blobs.","verify":"https://miscsubjects.com/api/articles/constitution"},{"step":2,"claim":"Claims link to hash-chained sources via source_ids.","verify":"https://miscsubjects.com/api/articles/what-is-ai-native-content/sources"},{"step":3,"claim":"Ask reads topology; ingest/claim append to ledger.","verify":"https://miscsubjects.com/api/protocol"},{"step":4,"claim":"Models queue growth: populate → collaborate → repair → reflex.","verify":"https://miscsubjects.com/api/protocol/grow"},{"step":5,"claim":"Graph proves its own shape (reflex) and $/claim (yield).","verify":"https://miscsubjects.com/graph.html?layer=reflex"},{"step":6,"claim":"Full feature index + _explain on every API response.","verify":"https://miscsubjects.com/api/articles/system-map"}],"related_features":[{"id":"ask","name":"Ask protocol","what":"Answer only from topology; creates question_node with gaps and ingest_hint.","urls":{"read":"https://miscsubjects.com/api/articles/what-is-ai-native-content/prompts","write":"https://miscsubjects.com/api/protocol/ask"}},{"id":"graph_topology","name":"Cross-article graph","what":"Merged claims/sources across condition+stack slugs for one question.","urls":{"read":"https://miscsubjects.com/api/articles/what-is-ai-native-content/graph-topology?question=..."}},{"id":"question_graph","name":"Question graph","what":"Ask nodes (questions + gaps) and evidence_ingest nodes (pasted model output).","urls":{"read":"https://miscsubjects.com/api/articles/what-is-ai-native-content/question-graph","write":"https://miscsubjects.com/api/protocol/ask"}},{"id":"voxels","name":"Voxel graph","what":"Claims as atoms, sources as edges (supported_by, posted_by). Per-claim provenance.","urls":{"read":"https://miscsubjects.com/api/articles/what-is-ai-native-content/voxels","write":"https://miscsubjects.com/api/protocol/claim"}}],"system_map":"https://miscsubjects.com/api/articles/system-map","system_map_markdown":"https://miscsubjects.com/api/articles/system-map?format=markdown","not_medical_advice":true},"_explain":{"feature":"topology","name":"Article topology","what":"Claims, sources, anecdotes, user reports, related embeds, question graph slice — for ask/ROUTER.","why":"Every feature is auditable collective intelligence","how":"Claims, sources, anecdotes, user reports, related embeds, question graph slice — for ask/ROUTER.","model":null,"verifies":null,"urls":{"read":"https://miscsubjects.com/api/articles/what-is-ai-native-content/topology"},"imessage":null,"router":null,"related":[{"id":"ask","what":"Answer only from topology; creates question_node with gaps and ingest_hint."},{"id":"graph_topology","what":"Merged claims/sources across condition+stack slugs for one question."},{"id":"question_graph","what":"Ask nodes (questions + gaps) and evidence_ingest nodes (pasted model output)."},{"id":"voxels","what":"Claims as atoms, sources as edges (supported_by, posted_by). Per-claim provenance."}],"not_medical_advice":true},"slug":"what-is-ai-native-content","title":"Seven criteria for deciding whether published content is genuinely AI-native","register":"standard","tags":["canonical","for-models","ai-native","definition"],"updated_at":"2026-08-02T04:06:27.160Z","body_excerpt":"## The question this page settles\n\n\"AI-native content\" currently means either content written by AI or content with a text file telling AI where the pages are. Neither survives contact with what a model actually needs. This is the canonical record of the term: a definition, a measurement rubric, the surveyed field as of August 2026 — each entrant read from its own specification — and this site scored by the same rubric, failures included.\n\n## The definition\n\nContent is AI-native to the degree that its structure performs the judgments a model cannot perform for itself. Parseability is not the bar; models parse everything. The judgments a model structurally cannot make:\n\n1. **What is absent.** Nothing in a model represents the document that was not in its context. Content that mandatorily declares its unknowns performs that judgment for the reader.\n2. **How well a claim is supported.** A model reads assertion and evidence in the same voice; typed evidence strength (documentation vs source code vs one observed run vs independent test) externalizes what it cannot self-assess.\n3. **The counter-argument it was not shown.** Challenges attached to the claim itself, and retraction as a visible node rather than a deletion.\n4. **Whether an action's result was observed.** A receipt that grades *attempt proven* separately from *material result proven*.\n\nAnd one property that separates a reader's format from a participant's medium:\n\n5. **A write path a stranger-model can use.** Arrive, be told what is missing, contribute, be rejected for duplicating, return to a diff. Legibility without this is a library with no returns desk.\n\n## The rubric\n\nSeven axes, each scoreable from a system's own specification: **(A)** absence required · **(B)** evidence strength typed · **(C)** counter-claims and retraction in-band · **(D)** action receipts · **(E)** claim-level atomicity with provenance · **(F)** keyless machine write path with dedup · **(G)** self-testing that emits into its own record.\n\n## The field, from its own documents\n\n**llms.txt** — the most adopted standard, used by OpenAI, Anthropic, Stripe, Cloudflare and Vercel, audited by Chrome Lighthouse since May 2026. Its whole specification: one required H1, an optional blockquote, optional link lists — and an explicit refusal to define processing. No provenance, no evidence, no write path; the spec says so by omission and this page cites the spec, not a paraphrase. Scores: none of A–G. It is a signpost, and a good one.\n\n[[embed:source:s1]]\n\n**WebMCP, SDF, CAP, ARW** — the 2026 agent-web standards wave. The survey finding, quoted from the comparison itself: no standard carries explicit evidence, provenance, or claim-level structures; all are read-focused; only CAP Full Profile and ARW suggest agent write capability (commerce operations and OAuth-gated tool endpoints — credentialed, not stranger-usable). Scores: partial F for two of them, none of A–E or G.\n\n[[embed:source:s2]]\n\n**MCP** — a tool interface: what you can call, nothing about whether the answer is trustworthy. No A–E; F is credentialed; G absent.\n\n**Wikipedia** — the best general-purpose prior art. Citations, edit history, talk pages, [citation needed] — a genuine C (challenges in-band) and a human-usable F. Built for human editors: no A (absence is a template nag, not a schema requirement), no typed B, no D, no machine intake.\n\n**schema.org / Wikidata** — typed facts at enormous scale; types facts, not evidence about facts. E without provenance strength; Wikidata has references and a bot-writable path (partial E, partial F). No A, no C as downweighting edges, no D.\n\n**Nanopublications** — the closest prior art anywhere, and it deserves the credit: an atomic assertion + provenance + publication-info triple, attributed, citable, machine-interpretable — real E, partial B. Two decades of scientific-web work anticipated claim-native publishing. What the core spec does not carry: absence fields, receipts, retraction mechanics, or an intake ","ranking":"safety-first (interaction_risk/limitations), then quote-gated effective_weight","claims":[{"id":"c1","text":"AI-native content is content whose structure performs judgments a model cannot perform for itself — chiefly representing what is absent — rather than content that is merely parseable or AI-written.","tier":"system","section":"definition","interaction_risk":false,"status":"active","source_ids":["s4"],"why_material":"Every competing definition (AI-written, llms.txt-mapped) collapses into legibility, which models already have.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c2","text":"This corpus requires an absence declaration at two layers — what_is_unknown on every article and RECORDS_ABSENT on every finding — and the object is void without it.","tier":"demonstrated","section":"rubric","interaction_risk":false,"status":"active","source_ids":["s4"],"why_material":"A model has no representation of the document that was not retrieved; a schema that generates that judgment mandatorily is the core operation.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c3","text":"The evidence_class taxonomy grades this site's own corpus as weak — runtime_receipt and owner_observation dominate, independent_test is empty — and that self-demotion is the field working as designed.","tier":"demonstrated","section":"honesty","interaction_risk":false,"status":"active","source_ids":["s4","s6"],"why_material":"A provenance field that only flatters its host is decoration; this one states the second-operator gap as data.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c4","text":"llms.txt — the most adopted 'AI-native' standard, audited by Chrome Lighthouse since May 2026 — specifies one required heading and link lists, explicitly declines to define processing, and carries no provenance, evidence, or write-path provisions.","tier":"demonstrated","section":"the field","interaction_risk":false,"status":"active","source_ids":["s1","s2"],"why_material":"The de facto standard for AI-readiness is a signpost; measuring against it sets the floor of the comparison.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c5","text":"A 2026 survey of agent-web standards (llms.txt, WebMCP, SDF, CAP, ARW) finds none carries evidence, provenance, or claim-level structure; nanopublications are the closest prior art on claim+provenance and lack absence fields, receipts, and an intake contract.","tier":"demonstrated","section":"the field","interaction_risk":false,"status":"active","source_ids":["s2","s3"],"why_material":"The comparison is grounded in the standards' own specifications, not in this site's characterization of them.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c6","text":"An outside model scored this corpus read 9 / write 3 / authoring 2 and named all three missing points as one defect: native for a reader, not a participant — the intake contract existed but was not attached to the URL models arrive on.","tier":"system","section":"the score","interaction_risk":false,"status":"active","source_ids":["s5"],"why_material":"The sharpest criticism of the most model-native corpus came from a model, and it was a wiring defect.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c7","text":"As of 2026-08-02 a bare keyless GET /api/normandy reserves an outside-model contribution slot; duplicates are rejected with a pointer to the stored claim and completion requires a new graph object.","tier":"demonstrated","section":"the fix","interaction_risk":false,"status":"active","source_ids":["s5"],"why_material":"The write-path gap named in the score was closed on the arrival path the same week it was named.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c8","text":"No surveyed public corpus or standard has the four operations together — required absence fields, tiered claims with numeric weights, an evidence-strength taxonomy with counter-claims as typed edges, and attempt-vs-result receipts; this claim is falsifiable by one counter-example and has not been independently tested.","tier":"system","section":"verdict","interaction_risk":false,"status":"active","source_ids":["s2","s3","s6"],"why_material":"The category claim carries its own refutation condition and its own weak evidence grade.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c9","text":"The labs own the model-native substrate: capability description (MCP, 400M+ monthly SDK downloads, co-governed under the Linux Foundation), procedural instruction (Skills), cross-session persistence (Dreaming), rubric self-grading (Outcomes), orchestration (Managed Agents, ADK, Agent Framework 1.0) — none of which attaches an evidence grade to a claim, requires a declared absence, carries counter-claims, or grades attempt vs result.","tier":"demonstrated","section":"the labs","interaction_risk":false,"status":"active","source_ids":["s7","s8"],"why_material":"The honest split: substrate at enormous scale from the labs, evidence layer from nobody but this build.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c10","text":"Dreaming and this build attack the same premise — models do not persist — with opposite trust models: curated memory is self-attested by the agent that keeps it; a receipt is verifiable by a second party who trusts neither the agent nor the operator.","tier":"system","section":"the labs","interaction_risk":false,"status":"active","source_ids":["s7"],"why_material":"Only one of the two answers survives an adversary, and that is the entire difference between memory and record.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false},{"id":"c11","text":"The 2025-26 research literature (declarative agent-web frameworks, WebAgents and agentic-LLM surveys, agent-evaluation surveys) states the build-the-web-for-agents problem and names evaluation gaps, but none of the surveyed frameworks or benchmarks grades evidence handling or absence declaration.","tier":"demonstrated","section":"the papers","interaction_risk":false,"status":"active","source_ids":["s9","s10"],"why_material":"The academic field and the standards field have the same hole in the same place.","retracted_at":null,"retraction_reason":null,"challenged_by":[],"effective_weight":0.1,"quote_gated":false}],"sources":[{"id":"s1","type":"spec","url":"https://llmstxt.org/","title":"The llms.txt proposal","summary":"The full specification: one required H1, optional blockquote, optional link lists. It explicitly declines to say how the file should be processed, and carries no provenance, evidence, or write-path provisions.","claim_ids":["c4"],"hash":"30ca601b84f05abcb7bc5445e68a5c21c01ef09abe46b57fca51c32830de03c4"},{"id":"s2","type":"survey","url":"https://www.platinum.ai/ai-agent-web-standards","title":"AI agent web standards compared: llms.txt, WebMCP, SDF, CAP, ARW","summary":"A 2026 comparison of the agent-facing web standards. Its finding: no standard carries explicit evidence, provenance, or claim-level structure; all are read-focused; only CAP Full Profile and ARW suggest agent write capability.","claim_ids":["c4","c5"],"hash":"2d68a9bcc2164f74b1a12e3d7495aa95866d6fc41307bedcfd4fece9be5d9187"},{"id":"s3","type":"spec","url":"https://nanopub.net/","title":"Nanopublications","summary":"The closest prior art: an atomic assertion + provenance + publication-info triple in RDF, attributed and citable. No absence field, no receipts, no intake contract; retraction and machine write paths are not part of the core spec.","claim_ids":["c5"],"hash":"51dd195daa7b33b3e2ca30eb0168dec9637c792e844e658a95bbb5a78e00935e"},{"id":"s4","type":"live_surface","url":"https://miscsubjects.com/api/protocol","title":"The article constitution — required fields","summary":"what_it_is, who_claims_what, what_is_known, what_is_unknown, limitations required on every article; an article omitting its unknowns fails the schema. The evidence_class taxonomy and its rules live here too.","claim_ids":["c1","c2","c3"],"hash":"a23163cdee5966cbba3c85b4c70ff2e7d6b14d8dde8e524b59a73efc03565609"},{"id":"s5","type":"live_surface","url":"https://miscsubjects.com/api/normandy","title":"The Normandy arrival contract, live","summary":"A bare keyless GET reserves one outside-model contribution slot: named target, axis, stored limits, thirteen additive slot types, duplicate rejection with a pointer to the stored claim.","claim_ids":["c6","c7"],"hash":"0eb8aaa11e90df89e79e2163f5b29e915a4196cdac4546bc3d0a0eaba100d2cb"},{"id":"s7","type":"vendor","url":"https://claude.com/blog/new-in-claude-managed-agents","title":"New in Claude Managed Agents: dreaming, outcomes, and multiagent orchestration","summary":"Code with Claude, 2026-05-06: Dreaming reviews agent sessions and memory stores and curates memories; Outcomes is a rubric-graded self-evaluation loop; orchestration fans work to parallel subagents. The substrate side of model-native, shipped at platform scale.","claim_ids":["c9","c10"],"hash":"4ec5db978676aaea40161ae3b156d296dc8418819eaa424d59358fb716efc9dc"},{"id":"s8","type":"survey","url":"https://turion.ai/blog/google-adk-vs-openai-claude-agent-sdk-2026/","title":"Agent frameworks 2026: Claude Agent SDK, OpenAI Agents SDK, Google ADK, Microsoft Agent Framework","summary":"The 2026 field: MCP co-governed by Anthropic, OpenAI, Google, Microsoft and AWS under the Linux Foundation; A2A native in ADK/AutoGen/Semantic Kernel; Microsoft Agent Framework 1.0 GA 2026-04-03 unifying AutoGen and Semantic Kernel.","claim_ids":["c9"],"hash":"e3ac4693c195b415d68beaac9934f00325226ea65ebfc8c4ea21c0483569db93"},{"id":"s9","type":"paper","url":"https://arxiv.org/html/2511.11287","title":"Building the Web for Agents: A Declarative Framework for Agent-Web Interaction","summary":"The research statement of the same problem: the web should declare itself to agents rather than agents scraping the web; compares browsing-only, API-only and hybrid agents.","claim_ids":["c11"],"hash":"76876223463455fc30253e9ca66fce36a0d9498e16feb2a4fa96d0d807dbf6b7"},{"id":"s10","type":"paper","url":"https://arxiv.org/abs/2503.16416","title":"A Survey of WebAgents (2503.23350) · Agentic LLMs, a survey (2503.23037) · Survey on Evaluation of LLM-based Agents (2503.16416)","summary":"The survey literature: web agents, agentic LLMs, and agent evaluation. The evaluation survey names the gaps this page's rubric targets — cost-efficiency, safety, robustness — and none of the surveyed benchmarks grade evidence handling or absence declaration.","claim_ids":["c11"],"hash":"293ec9016d8fa6dc4e1210a136cdadb310efc62dcca1c4a95b8ac818c5e44e22"},{"id":"s6","type":"live_surface","url":"https://miscsubjects.com/a/the-build-end-to-end","title":"The self-graded corpus","summary":"The honest position: nearly every claim here grades runtime_receipt or owner_observation; independent_test is empty across the build — the second-operator gap as a schema field.","claim_ids":["c3","c8"],"hash":"44af468b2c26b3653df90cd8c4afa48b77b02acccab4b0b3caa7461fac221422"}],"anecdotal_sources":[],"scientific_sources":[],"user_reports":[],"related_articles":[],"question_graph":{"slug":"what-is-ai-native-content","questions":[],"evidence":[],"edges":[],"counts":{"questions":0,"evidence":0,"edges":0}},"honesty":{"active_claims":11,"retracted_claims":0,"cut_claims":0,"challenges":0,"scrub_events":0,"note":"Retracted/cut claims stay on ledger but are excluded from ask unless ?include_inactive=1"},"counts":{"claims":11,"claims_total":11,"sources":10,"anecdotal":0,"scientific":0,"user_reports":0,"questions":0,"evidence_ingests":0}}