Em 0036

Source content

{

"id": "EM-0036",

"title": "Admit Cases 003 and 004 into the four-case How We Know library",

"status": "ready",

"change_class": "ordinary-implementation",

"objective": "Select the exact independently reviewed EM-0034 and EM-0035 dossiers through accepted manifests, extend the reversible application dossier library with a generic bounded-proposition profile, and expose four coherent How We Know cases across human and agent interfaces without changing accepted evidence meaning or claiming deployment before provider read-back.",

"depends_on": [

"EM-0030",

"EM-0034",

"EM-0035"

],

"authority": {

"allowed_paths": [

"README.md",

"catalog/**",

"src/**",

"tests/**",

"docs/**",

"ops/**",

"runs/**"

],

"forbidden_paths": [

"constitution/**",

"policies/**",

"schemas/**",

"governance/events/**",

".github/**",

"tasks/contracts/**",

"research/**"

]

},

"required_evaluation": [

"exact dossier and independent-review-receipt byte, digest, length, format, reviewer, reviewed-head, reviewed-tree, and independence binding for Cases 003 and 004",

"deterministic manifest enumeration with duplicate number, slug, dossier identity, profile, and route rejection",

"byte and semantic regression protection for every accepted Case 001 and Case 002 dossier, receipt, manifest, route, policy view, count, and generated representation",

"generic reversible bounded-proposition adapter tests that preserve each dossier's own count grammar, terminology, lineage dimensions, unresolved objects, and practical reading",

"relation-derived scoreboards, complete count-to-ledger navigation, sentence-to-source closure, and unresolved or no-credit visibility for Cases 003 and 004",

"material encyclopedia-versus-skeptical output-difference tests over identical evidence objects for both new cases",

"four-case How We Know index, homepage discovery, review receipts, Markdown, JSON, share cards, llms.txt, sitemap, OpenAPI, local REST, MCP, and CLI parity tests",

"cold-start agent discovery and bounded summary tests using only committed public outputs for all four cases",

"no-JavaScript fallback, native keyboard accessibility, desktop and mobile visual review, link integrity, content-type, and canonical-URL checks",

"whole-release catalog, frontier, compiler, accepted-commit, content-digest, manifest, and per-case dossier identity-parity tests",

"public disclosure noninterference audit, full deterministic make check, exact path-scope comparison, and clean tracked state",

"fresh-clone independent exact-head review before protected merge",

"provider and route read-back after a separately authorized accepted Pages deployment"

],

"acceptance": [

"explicit accepted manifests select exactly the independently reviewed Case 003 and Case 004 dossiers and receipts without modifying their bytes",

"the compiler discovers exactly four accepted dossier manifests deterministically and fails closed on duplicate identities, numbers, slugs, unsupported profiles or formats, receipt drift, or route collisions",

"Cases 001 and 002 remain byte-bound, fully reachable, and semantically unchanged while the How We Know index lists four distinct evidence files",

"Cases 003 and 004 each have stable encyclopedia, skeptical, review, Markdown, JSON, share-card, API-style, MCP, and CLI representations derived from one disclosure-safe dossier",

"the generic application profile renders case-specific count labels and ledgers without forcing either new case into Case 001's replication grammar or Case 002's agent-citation grammar",

"every displayed count opens a complete ledger and every material sentence exposes the exact accepted work, edition, span, digest, calculation, dependence relation, and license treatment required for inspection",

"the skeptical view changes the first-screen finding and practical reading for each new case rather than merely changing a path or style",

"the homepage and How We Know index present a coherent four-case library without advertising unreviewed future cases or describing the site as broadly launched",

"a cold-start agent can discover all four cases from llms.txt and recover each question, bounded verdict, count grammar, dependence warning, unresolved set, and review limitations",

"all emitted interfaces carry one coherent catalog, frontier, accepted commit, compiler, policies, dossier identities, and content digest for the release",

"make check passes without changing accepted source state",

"after separately authorized deployment, epistemedia.org read-back confirms every new case route, content type, canonical URL, release identity, keyboard path, mobile layout, and four-case discovery surface before broad promotion"

],

"limitations": [

"This task admits two bounded historical evidence files; it does not establish a universal evaluation schema, current model ranking, general law of communication, or systematic-review completeness.",

"The application dossier and manifest formats remain reversible alpha adapters and are not promoted into normative schemas or protocol semantics.",

"Cases 001 and 002 research meaning and accepted bytes must not change as a side effect of adding the library entries.",

"A merged configuration or successful local build is not proof of live deployment; production status requires a separate authorized Pages run and provider plus route read-back.",

"Broad launch or promotion remains gated until all four cases are live, coherent, independently reviewed, and provider-read-back; hosted API or MCP activation, new lenses, Case 005 research, workflow privilege changes, DNS changes, package publication, provider calls, credentials, accounts, and spend remain out of scope."

]

}