@inflexa-ai/harness 0.17.0 → 0.19.0

package/dist/contracts/target-dossier.js

@@ -7,18 +7,30 @@
  *
  * Coverage discipline: every section that depends on enrichment data
  * carries a `coverage` discriminator (`available | queried_no_data |
- * not_loaded`). Sections that fall back to inference also carry an
- * `inference_path`. Aggregate rows preserve their contributing evidence
- * under `evidence: [...]` arrays.
- *
- * `DossierSchema` is a single v3 shape — no backwards-compatibility union.
- * Old v1/v2 persisted rows will not parse; that is intentional.
+ * not_loaded | filtered`), taken from the shared envelope rather than
+ * declared per section. `filtered` says a threshold of ours emptied the
+ * section and carries the filter and the count it discarded; a partially
+ * filtered `available` section carries `dropped_count`. Sections that fall
+ * back to inference also carry an `inference_path`. Aggregate rows preserve
+ * their contributing evidence under `evidence: [...]` arrays; rows that
+ * assert a liability carry it under `support`.
  *
  * See `TARGET_DOSSIER.md` for the full editorial specification.
  */
 import { z } from "zod";
+import { OrganSystemSchema } from "./organ-system.js";
+import { SeveritySchema } from "./severity.js";
 // ── Coverage primitives ─────────────────────────────────────────────
-export const CoverageSchema = z.enum(["available", "queried_no_data", "not_loaded"]);
+export const CoverageSchema = z.enum(["available", "queried_no_data", "not_loaded", "filtered"]);
+/**
+ * Coverage states a per-row marker can take.
+ *
+ * `filtered` is a section-level outcome only: a filter that removes a row
+ * removes the row, so a row that survived to be read can never report it.
+ * Derived from the section vocabulary rather than restated, so the two
+ * cannot drift into separate lists.
+ */
+export const RowCoverageSchema = CoverageSchema.exclude(["filtered"]);
 const CoverageError = z.object({
     message: z.string(),
     source: z.string().optional(),
@@ -27,12 +39,18 @@ const CoverageError = z.object({
  * Build a discriminated section schema: when `coverage` is `available`,
  * `data` is required and matches the inner schema. Otherwise `data` is
  * absent and an optional `error` payload may be present.
+ *
+ * `dropped_count` rides on the `available` branch as well as the `filtered`
+ * one because a filter that removes some rows still leaves a section whose
+ * list is partial — without it, a section that discarded most of its rows
+ * reports a clean `available` and overstates its own completeness.
  */
 function withCoverage(data) {
     return z.discriminatedUnion("coverage", [
         z.object({
             coverage: z.literal("available"),
             data,
+            dropped_count: z.number().int().nonnegative().optional(),
             inference_path: z.string().optional(),
             truncated: z.boolean().optional(),
         }),
@@ -41,6 +59,12 @@ function withCoverage(data) {
             error: CoverageError.optional(),
             inference_path: z.string().optional(),
         }),
+        z.object({
+            coverage: z.literal("filtered"),
+            filter: z.string(),
+            dropped_count: z.number().int().nonnegative(),
+            inference_path: z.string().optional(),
+        }),
         z.object({
             coverage: z.literal("not_loaded"),
             reason: z.string().optional(),
@@ -61,6 +85,7 @@ export const SECTION_BLURBS = {
 export const EvidenceItemSchema = z.object({
     pmid: z.string().optional(),
     doi: z.string().optional(),
+    accession: z.string().optional(),
     source: z.string(),
     predicate: z.string().optional(),
     score: z.number().optional(),
@@ -79,15 +104,58 @@ export const EvidenceItemSchema = z.object({
         .optional(),
 });
 const EvidenceList = z.array(EvidenceItemSchema);
+/**
+ * Evidence attached to a claim, which must resolve to something a reader can
+ * actually go and check. `source` alone names a provenance bucket, not a
+ * record, so it does not qualify on its own.
+ *
+ * The constraint lives here rather than on `EvidenceItemSchema` because that
+ * shape is shared with sections whose producers legitimately carry looser
+ * evidence; tightening it globally would break them, and loosening the claim
+ * to match them would defeat the point.
+ */
+export const ClaimEvidenceSchema = EvidenceItemSchema.refine((e) => Boolean(e.pmid ?? e.doi ?? e.accession ?? e.regulatory_reference), {
+    message: "claim evidence needs a locator: pmid, doi, accession, or regulatory_reference",
+});
+/**
+ * What backs a claim: either evidence, or a stated reason there is none.
+ *
+ * Discriminating on `state` — rather than making `evidence` nullable and
+ * checking it at runtime — is what makes a scored claim with no evidence
+ * unconstructable rather than merely invalid. `unknown` is a complete,
+ * first-class outcome: an invariant that is harder to satisfy honestly than
+ * dishonestly makes the data worse, not better.
+ */
+export const ClaimSupportSchema = z.discriminatedUnion("state", [
+    z.object({
+        state: z.literal("scored"),
+        evidence: z.array(ClaimEvidenceSchema).min(1),
+    }),
+    z.object({
+        state: z.literal("unknown"),
+        reason: z.string().min(1),
+    }),
+]);
+/**
+ * Attach support to a claim body.
+ *
+ * The support rides as a field rather than wrapping the body, because these
+ * rows *are* the assertion: a reader of an unevidenced liability still needs
+ * the liability. Swapping the body out on the unknown branch would leave
+ * nothing to read.
+ */
+function withEvidence(body) {
+    return body.extend({ support: ClaimSupportSchema });
+}
 // ── §1.1 Entity ─────────────────────────────────────────────────────
 export const EntitySchema = z.object({
     canonicalId: z.string(),
     symbol: z.string(),
     displayName: z.string(),
     /**
-     * Cortex's Phase-0 resolver currently only emits `"gene"`. The schema
-     * keeps the broader enum so the contract is forward-compatible with
-     * protein-/rna-/complex-level entities once the resolver supports them.
+     * Cortex's Phase-0 resolver only emits `"gene"`. The schema keeps the
+     * broader enum so protein-, rna- and complex-level entities are
+     * expressible once the resolver supports them.
      */
     entityType: z.enum(["gene", "protein", "rna", "complex"]),
     ontology: z.string(),
@@ -101,46 +169,46 @@ export const EntitySchema = z.object({
     synonyms: z.array(z.string()).default([]),
     proteinFamily: z.string().optional(),
 });
-// ── §1.2 Summary ────────────────────────────────────────────────────
-export const SummarySchema = z.object({
-    total_evidence_items: z.number().int().nonnegative(),
-    total_distinct_papers: z.number().int().nonnegative(),
-    has_human_evidence: z.boolean(),
-    has_clinical_evidence: z.boolean(),
-    strongest_strength_label: z.string().nullable(),
-    highest_score: z.number().nullable(),
-    highest_score_is_conflicted: z.boolean(),
-    total_distinct_clinical_trials: z.number().int().nonnegative(),
-});
 // ── §2.1 Liability summary ──────────────────────────────────────────
-export const LiabilityBulletSchema = z.object({
+/**
+ * The assertion a liability bullet makes, without its support.
+ *
+ * Named separately because a dropped bullet records the same assertion
+ * plus the reason it was dropped, and carries no support of its own.
+ */
+export const LiabilityBulletBodySchema = z.object({
     text: z.string(),
     rationale: z.string(),
     category: z.enum(["fatal_post_market", "class_liability", "off_target_safety", "high_safety_organ_expression", "broad_expression", "other"]),
 });
-export const LiabilitySummarySchema = z.object({
+export const LiabilityBulletSchema = withEvidence(LiabilityBulletBodySchema);
+/**
+ * Liability bullets removed post-synthesis because they cited a PMID the
+ * direction-of-effect auditor found contradicts the bullet's framing.
+ * Mirrors `executive_recommendation.key_risks_dropped` — the dropped
+ * content is preserved with the reason for an audit trail.
+ */
+export const LiabilityBulletDroppedSchema = LiabilityBulletBodySchema.extend({
+    reason: z.string(),
+    cited_pmid: z.string().optional(),
+});
+/** `.strict()` rejects the derived counters, which live in the derived sub-tree. */
+export const LiabilitySummarySchema = z
+    .object({
     liability_bullets: z.array(LiabilityBulletSchema),
+    liability_bullets_dropped: z.array(LiabilityBulletDroppedSchema).optional(),
     modality_recommendation: z.string().nullable(),
-    expression_breadth: z.object({
-        high_expression_tissue_count: z.number().int().nonnegative(),
-        total_assessed_tissues: z.number().int().nonnegative(),
-    }),
-    class_liability_count: z.number().int().nonnegative(),
-    safety_target_off_target_count: z.number().int().nonnegative(),
-    off_tissue_risk_organ_count: z.number().int().nonnegative(),
     same_class_drug_count: z.number().int().nonnegative(),
-    safety_data_sources_checked: z.number().int().nonnegative(),
     inferred_therapeutic_area: z.string().nullable(),
     no_liabilities_disclosure: z.string().optional(),
-});
+})
+    .strict();
 // ── §2.2 Tractability ───────────────────────────────────────────────
 export const TractabilityModalitySchema = z.object({
     /**
-     * Phase-4 currently emits only `small_molecule`, `antibody`, and
-     * `other_clinical`. The remaining variants are reserved for future
-     * Phase-2 modality-fallback expansions (PROTAC, oligonucleotide, peptide
-     * libraries) — keep them in the enum so persisted dossiers don't reject
-     * once those land.
+     * Phase-4 emits `small_molecule`, `antibody`, and `other_clinical`. The
+     * remaining variants belong to the Phase-2 modality fallback (PROTAC,
+     * oligonucleotide, peptide libraries).
      */
     modality: z.enum(["small_molecule", "antibody", "protac", "other_clinical", "oligonucleotide", "peptide"]),
     levels: z.array(z.string()),
@@ -150,9 +218,31 @@ export const TractabilityModalitySchema = z.object({
     approved_drug_ids: z.array(z.string()).optional(),
     note: z.string().optional(),
 });
-export const TractabilitySchema = withCoverage(z.object({
+const MOLECULE_TYPE_TO_MODALITY = {
+    Protein: ["peptide", "antibody"],
+    "Small molecule": ["small_molecule"],
+    Oligonucleotide: ["oligonucleotide"],
+};
+export const TractabilitySchema = withCoverage(z
+    .object({
     modalities: z.array(TractabilityModalitySchema),
     preferred_modality: z.string().nullable(),
+    drug_molecule_types_present: z.array(z.string()).default([]),
+})
+    .superRefine((v, ctx) => {
+    for (const moleculeType of v.drug_molecule_types_present) {
+        const allowed = MOLECULE_TYPE_TO_MODALITY[moleculeType];
+        if (!allowed)
+            continue; // unknown molecule_type — skip rather than block
+        const hasMatching = v.modalities.some((m) => allowed.includes(m.modality) && !m.is_inferred_from_family);
+        if (!hasMatching) {
+            ctx.addIssue({
+                code: "custom",
+                message: `drug molecule_type "${moleculeType}" has no matching enumerated modality row`,
+                path: ["modalities"],
+            });
+        }
+    }
 }));
 // ── §2.3 Indications ────────────────────────────────────────────────
 export const IndicationRowSchema = z.object({
@@ -197,6 +287,62 @@ export const DrugInteractionRowSchema = z.object({
 });
 export const DrugInteractionsSchema = withCoverage(z.object({ rows: z.array(DrugInteractionRowSchema) }));
 // ── §2.5 Clinical development ───────────────────────────────────────
+export const ClinicalTrialAttributionBasisSchema = z.object({
+    kind: z.enum([
+        "mechanism_target_match",
+        "known_class_drug",
+        "therapeutic_program_match",
+        "related_family_target",
+        "biomarker_endpoint",
+        "condition_only",
+        "text_match",
+        "manual",
+    ]),
+    source: z.string(),
+    excerpt: z.string().optional(),
+    metadata: z.record(z.string(), z.unknown()).optional(),
+});
+export const ResolvedTrialInterventionSchema = z.object({
+    name: z.string(),
+    intervention_type: z.string().nullable().optional(),
+    chembl_id: z.string().nullable().optional(),
+    therapeutic_program_id: z.string().nullable().optional(),
+    target_uniprots: z.array(z.string()).default([]),
+    resolver_source: z.string(),
+});
+export const ClinicalTrialAttributionSchema = z
+    .object({
+    relationship: z.enum([
+        "direct_modulator",
+        "class_modulator",
+        "related_family_target",
+        "target_biomarker",
+        "pathway_biomarker",
+        "condition_only",
+        "unrelated",
+        "unknown",
+    ]),
+    evidence_role: z.enum(["supports_target", "contextual", "excluded"]),
+    basis: z.array(ClinicalTrialAttributionBasisSchema).min(1),
+    resolved_interventions: z.array(ResolvedTrialInterventionSchema).default([]),
+    exclusion_reason: z.string().optional(),
+})
+    .superRefine((v, ctx) => {
+    if (v.evidence_role !== "supports_target" && !v.exclusion_reason) {
+        ctx.addIssue({
+            code: "custom",
+            message: "contextual/excluded trial evidence requires exclusion_reason",
+            path: ["exclusion_reason"],
+        });
+    }
+    if (v.evidence_role === "supports_target" && v.relationship !== "direct_modulator" && v.relationship !== "class_modulator") {
+        ctx.addIssue({
+            code: "custom",
+            message: "supports_target evidence must be direct_modulator or class_modulator",
+            path: ["relationship"],
+        });
+    }
+});
 export const ClinicalTrialRowSchema = z.object({
     nct_id: z.string(),
     title: z.string(),
@@ -206,34 +352,51 @@ export const ClinicalTrialRowSchema = z.object({
     start_date: z.string().nullable(),
     completion_date: z.string().nullable(),
     match_confidence: z.enum(["high", "medium", "low", "off_target"]),
+    attribution: ClinicalTrialAttributionSchema,
+    eligible_for_toxicology_aggregation: z.boolean(),
 });
 export const TrialOutcomeRowSchema = z.object({
     nct_id: z.string(),
-    outcome_type: z.enum(["primary", "secondary", "serious_ae", "non_serious_ae", "other_ae"]),
     measure: z.string(),
+    outcome_type: z.enum(["primary", "secondary", "other"]),
     description: z.string().optional(),
-    result: z.string().optional(),
-});
-export const FailedTrialRowSchema = z.object({
-    nct_id: z.string(),
-    title: z.string(),
-    why_stopped: z.string(),
-    failure_category: z.enum(["safety", "efficacy", "strategic", "other"]),
-    classifier: z.enum(["rules", "llm"]),
+    time_frame: z.string().optional(),
+    effect: z.discriminatedUnion("kind", [
+        z.object({
+            kind: z.literal("quantitative"),
+            value: z.number(),
+            ci_low: z.number().optional(),
+            ci_high: z.number().optional(),
+            units: z.string(),
+        }),
+        z.object({
+            kind: z.literal("qualitative"),
+            direction: z.enum(["increase", "decrease", "no_change"]),
+            magnitude_label: z.string(),
+        }),
+        z.object({
+            kind: z.literal("not_extracted"),
+            reason: z.enum(["ctgov_no_numeric_result", "ctgov_no_result_groups", "parse_failure"]),
+        }),
+    ]),
+    attribution: ClinicalTrialAttributionSchema.optional(),
+    eligible_for_toxicology_aggregation: z.boolean().default(false),
 });
-const FailureCategoryV4Schema = z.discriminatedUnion("category", [
+const FailureCategorySchema = z.discriminatedUnion("category", [
     z.object({ category: z.literal("safety"), safety_evidence_excerpt: z.string() }),
     z.object({ category: z.literal("strategic"), category_evidence_excerpt: z.string() }),
     z.object({ category: z.literal("operational"), category_evidence_excerpt: z.string() }),
     z.object({ category: z.literal("efficacy"), category_evidence_excerpt: z.string() }),
 ]);
-export const FailedTrialRowV4Schema = z
+export const FailedTrialRowSchema = z
     .object({
     nct_id: z.string(),
     title: z.string(),
     why_stopped: z.string(),
     classifier: z.enum(["rules", "llm"]),
-    failure_category: FailureCategoryV4Schema,
+    failure_category: FailureCategorySchema,
+    attribution: ClinicalTrialAttributionSchema,
+    eligible_for_toxicology_aggregation: z.boolean(),
 })
     .superRefine((v, ctx) => {
     const excerpt = "safety_evidence_excerpt" in v.failure_category ? v.failure_category.safety_evidence_excerpt : v.failure_category.category_evidence_excerpt;
@@ -274,31 +437,25 @@ export const ClinicalBenchmarksSchema = z.object({
 export const ClinicalDevelopmentSchema = z.object({
     trials: withCoverage(z.object({
         rows: z.array(ClinicalTrialRowSchema),
-        related_target_trials: z
-            .array(ClinicalTrialRowSchema.extend({
-            match_confidence: z.literal("off_target"),
-        }))
-            .optional(),
-        related_receptor: z.string().optional(),
+        excluded_rows: z.array(ClinicalTrialRowSchema).default([]),
         selection_criteria: z
             .object({
-            derived_from: z.literal("analytics.discovery_trials"),
-            min_confidence: z.enum(["high", "medium"]),
-            excluded_off_target_count: z.number().int().nonnegative(),
+            derived_from: z.string().optional(),
+            min_confidence: z.string().optional(),
+            excluded_off_target_count: z.number().int().nonnegative().optional(),
         })
             .optional(),
     })),
     outcomes: withCoverage(z.object({ rows: z.array(TrialOutcomeRowSchema) })),
     failed_trials: withCoverage(z.object({
         rows: z.array(FailedTrialRowSchema),
-        related_target_trials: z.array(FailedTrialRowSchema).optional(),
-        related_receptor: z.string().optional(),
+        excluded_rows: z.array(FailedTrialRowSchema).default([]),
     })),
-    benchmarks: withCoverage(ClinicalBenchmarksSchema),
+    benchmarks: ClinicalBenchmarksSchema,
 });
 // ── §2.6 Safety profile ─────────────────────────────────────────────
 export const OrganRiskRowSchema = z.object({
-    organ: z.enum(["cardiac", "hepatic", "cns", "renal", "gi", "hematologic", "metabolic", "immune", "respiratory"]),
+    organ: OrganSystemSchema,
     risk_level: z.enum(["high", "medium", "low"]),
     signal_type_count: z.number().int().nonnegative(),
     signals: z.object({
@@ -309,34 +466,7 @@ export const OrganRiskRowSchema = z.object({
     }),
     evidence: EvidenceList,
 });
-export const FaersSummarySchema = z.object({
-    total_reports: z.number().int().nonnegative(),
-    fatal_report_count: z.number().int().nonnegative(),
-    any_fatal: z.boolean(),
-    seriousness_profile: z
-        .object({
-        death: z.number().int().nonnegative(),
-        hospitalization: z.number().int().nonnegative(),
-        life_threatening: z.number().int().nonnegative(),
-        disabling: z.number().int().nonnegative(),
-        congenital_anomaly: z.number().int().nonnegative(),
-        other_serious: z.number().int().nonnegative(),
-    })
-        .optional(),
-    top_signals: z.array(z.object({
-        meddra_term: z.string(),
-        organ: z.string().optional(),
-        report_count: z.number().int().nonnegative(),
-        proportional_reporting_ratio: z.number().optional(),
-    })),
-    per_modulator: z.array(z.object({
-        modulator: z.string(),
-        modulator_id: z.string().nullable(),
-        report_count: z.number().int().nonnegative(),
-        coverage: CoverageSchema,
-    })),
-});
-const SeriousnessAvailableV4Schema = z
+const SeriousnessAvailableSchema = z
     .object({
     coverage: z.literal("available"),
     total_reports: z.number().int().nonnegative(),
@@ -376,13 +506,13 @@ const SeriousnessAvailableV4Schema = z
         });
     }
 });
-const SeriousnessQueriedNoDataV4Schema = z.object({
+const SeriousnessQueriedNoDataSchema = z.object({
     coverage: z.literal("queried_no_data"),
     total_reports: z.number().int().nonnegative(),
 });
-export const FaersSummaryV4Schema = z.object({
+export const FaersSummarySchema = z.object({
     total_reports: z.number().int().nonnegative(),
-    seriousness: z.discriminatedUnion("coverage", [SeriousnessAvailableV4Schema, SeriousnessQueriedNoDataV4Schema]),
+    seriousness: z.discriminatedUnion("coverage", [SeriousnessAvailableSchema, SeriousnessQueriedNoDataSchema]),
     top_signals: z.array(z.object({
         meddra_term: z.string(),
         organ: z.string().optional(),
@@ -393,7 +523,7 @@ export const FaersSummaryV4Schema = z.object({
         modulator: z.string(),
         modulator_id: z.string().nullable(),
         report_count: z.number().int().nonnegative(),
-        coverage: CoverageSchema,
+        coverage: RowCoverageSchema,
     })),
 });
 export const TrialAesSchema = z.object({
@@ -409,11 +539,7 @@ export const TrialAesSchema = z.object({
         organ: z.string().optional(),
         nct_ids: z.array(z.string()),
     })),
-    /**
-     * Floor pct for non-serious AEs. PR-E lowered the operational value
-     * from 5 to 2 so real low-incidence AEs survive the filter; older
-     * persisted dossiers with the prior 5% floor still parse.
-     */
+    /** Floor pct for non-serious AEs, low enough to keep real low-incidence AEs. */
     non_serious_floor_pct: z.number().nonnegative().default(2),
     /**
      * Count of rows dropped by the catch-all-bucket filter (e.g., "Other"
@@ -429,28 +555,7 @@ export const TrialAesSchema = z.object({
     })
         .optional(),
 });
-export const OffTargetRowSchema = z.object({
-    off_target_id: z.string().nullable(),
-    off_target_name: z.string(),
-    target_class: z.string().optional(),
-    pchembl: z.number(),
-    is_safety_panel_target: z.boolean(),
-    organ_system: z.string().nullable(),
-    clinical_consequence: z.string().nullable(),
-    selectivity: z.object({
-        log_units: z.number(),
-        fold: z.number(),
-        legacy_alias: z.number().optional(),
-    }),
-    selectivity_window_below_threshold: z.boolean(),
-    evidence: EvidenceList,
-    metadata: z
-        .object({
-        merged_chembl_ids: z.array(z.string()).optional(),
-    })
-        .optional(),
-});
-const SelectivityV4Schema = z.union([
+const SelectivitySchema = z.union([
     z.object({
         vs_primary_potency: z.object({
             primary_pchembl_used: z.number(),
@@ -464,32 +569,39 @@ const SelectivityV4Schema = z.union([
         reason: z.string().min(1),
     }),
 ]);
-const OffTargetRowV4Base = z.object({
+/**
+ * The measured facts about an off-target hit, shared by the rows that
+ * assert a liability and the rows that record why one was set aside.
+ */
+export const OffTargetRowBodySchema = z.object({
     off_target_id: z.string().nullable(),
     off_target_name: z.string(),
     target_class: z.string().optional(),
     pchembl: z.number(),
     is_safety_panel_target: z.boolean(),
-    organ_system: z.string().nullable(),
+    organ_system: OrganSystemSchema.nullable(),
     clinical_consequence: z.string().nullable(),
-    selectivity: SelectivityV4Schema,
+    selectivity: SelectivitySchema,
     selectivity_window_below_threshold: z.boolean(),
-    evidence: EvidenceList,
     metadata: z.object({ merged_chembl_ids: z.array(z.string()).optional() }).optional(),
 });
-export const OffTargetRowV4Schema = OffTargetRowV4Base.extend({
-    relationship: z.literal("off_target"),
-});
-export const ExcludedOffTargetRowV4Schema = OffTargetRowV4Base.extend({
+/** The claim's own evidence rides in `support`, so the row carries no second evidence list. */
+export const OffTargetRowSchema = withEvidence(OffTargetRowBodySchema.extend({ relationship: z.literal("off_target") }));
+/**
+ * Excluded rows record why something was set aside rather than asserting a
+ * liability, so they carry evidence without the claim contract.
+ */
+export const ExcludedOffTargetRowSchema = OffTargetRowBodySchema.extend({
     relationship: z.enum(["intended_co_target", "on_target_self_hit", "obligate_cofactor"]),
     reason: z.string().min(1),
+    evidence: EvidenceList,
 });
-export const OffTargetPanelV4Schema = z.object({
-    rows: z.array(OffTargetRowV4Schema),
-    excluded_rows: z.array(ExcludedOffTargetRowV4Schema),
+export const OffTargetPanelSchema = z.object({
+    rows: z.array(OffTargetRowSchema),
+    excluded_rows: z.array(ExcludedOffTargetRowSchema),
 });
 export const ClassLiabilityOrganSchema = z.object({
-    organ: z.string(),
+    organ: OrganSystemSchema,
     drug_count_in_class: z.number().int().nonnegative(),
     drugs_with_signal: z.number().int().nonnegative(),
     signal_fraction: z.number(),
@@ -509,37 +621,196 @@ export const ClassPrecedentSchema = z.object({
     })),
     per_organ: z.array(ClassLiabilityOrganSchema),
 });
-export const SafetyFlagSchema = z.object({
-    organ: z.string(),
+export const SafetyFlagSchema = withEvidence(z.object({
+    organ: OrganSystemSchema,
     trail: z.string(),
     mechanism_hypothesis: z.string().nullable().optional(),
+    severity: SeveritySchema,
+}));
+export const RegulatoryActionRowSchema = z
+    .object({
+    drug_chembl_id: z.string(),
+    drug_name: z.string(),
+    agency: z.enum(["FDA", "EMA", "MHRA", "PMDA", "Health Canada", "TGA"]),
+    action_kind: z.enum(["referral", "withdrawal", "indication_restriction", "REMS", "black_box", "DHCP", "safety_communication", "label_warning"]),
+    action_date: z.string(),
+    finding: z.string(),
+    source_url: z.string().url().optional(),
+    source_kind: z.enum(["label_warning", "referral", "rems", "withdrawal", "boxed_warning", "safety_communication"]),
+    application_number: z.string().optional(),
+    label_section: z.string().optional(),
+    source_date: z.string().optional(),
+    evidence: EvidenceList,
+})
+    .superRefine((v, ctx) => {
+    if (v.source_kind === "label_warning" && v.action_kind === "safety_communication") {
+        ctx.addIssue({
+            code: "custom",
+            message: "FDA label warnings must not be represented as action_kind:safety_communication",
+            path: ["action_kind"],
+        });
+    }
 });
-export const RiskSummarySchema = z.object({
-    highest_risk_organ: z.string().nullable(),
-    off_target_safety_target_hits: z.number().int().nonnegative(),
-    class_liability_count: z.number().int().nonnegative(),
-    any_fatal_signal: z.boolean(),
-});
+/**
+ * One warning fragment from an FDA label, attributed to a canonical organ.
+ *
+ * The application number plus the label section are what let a reader open the
+ * label and find the excerpt again, which is why the row carries both rather
+ * than a bare drug name.
+ */
+export const RegulatoryOrganSignalRowSchema = withEvidence(z.object({
+    organ: OrganSystemSchema,
+    drug_name: z.string(),
+    agency: z.literal("FDA"),
+    application_number: z.string(),
+    label_section: z.string(),
+    excerpt: z.string(),
+}));
+export const RegulatoryOrganSignalsSchema = withCoverage(z.object({ rows: z.array(RegulatoryOrganSignalRowSchema) }));
 export const SafetyProfileSchema = z.object({
     organ_rollup: withCoverage(z.object({ rows: z.array(OrganRiskRowSchema) })),
     faers: withCoverage(FaersSummarySchema),
     trial_aes: withCoverage(TrialAesSchema),
-    off_target_panel: withCoverage(z.object({ rows: z.array(OffTargetRowSchema) })),
+    off_target_panel: withCoverage(OffTargetPanelSchema),
     failed_trials_safety_lens: withCoverage(z.object({
         rows: z.array(FailedTrialRowSchema),
-        related_target_trials: z.array(FailedTrialRowSchema).optional(),
-        related_receptor: z.string().optional(),
+        excluded_rows: z.array(FailedTrialRowSchema).default([]),
     })),
     class_precedent: withCoverage(ClassPrecedentSchema),
-    target_organ_liabilities: z.array(SafetyFlagSchema),
-    risk_summary: RiskSummarySchema,
+    target_organ_liabilities: withCoverage(z.object({ rows: z.array(SafetyFlagSchema) })),
+    regulatory_actions: withCoverage(z.object({ rows: z.array(RegulatoryActionRowSchema) })).optional(),
+    regulatory_organ_signals: RegulatoryOrganSignalsSchema.optional(),
+});
+// ── §2.6a Safety corroboration spine ────────────────────────────────
+/**
+ * One source's signal for one organ.
+ *
+ * The source names itself in a field rather than occupying a field of its own.
+ * That is what keeps the contributing set open: a source that did not exist
+ * when this shape was written contributes by emitting a record, with no schema
+ * edit, no new branch, and no consumer change. A struct with one field per
+ * source would instead force a schema change per source — and, because a new
+ * source arriving before its field exists has nowhere to land, would invite a
+ * second, narrower evidence path alongside this one.
+ *
+ * `evidence` is claim evidence, so a contribution always resolves to a record a
+ * reader can go and check.
+ */
+export const OrganSignalContributionSchema = z.object({
+    /** Stable id of the source that produced the signal. */
+    source: z.string().min(1),
+    /** What this source says about the organ, in the source's own terms. */
+    signal: z.string().min(1),
+    evidence: ClaimEvidenceSchema,
+});
+/**
+ * The corroboration record for one organ: which sources carry a signal for it,
+ * how many of them are independent, and what each contributed.
+ *
+ * `independent_source_count` counts distinct source ids, not contributions — a
+ * source carrying three label warnings for one organ is still one source. The
+ * claim's evidence is the contributions' evidence, which is why a signal with
+ * no locator is never admitted as a contribution in the first place.
+ */
+export const OrganCorroborationRowSchema = withEvidence(z.object({
+    organ: OrganSystemSchema,
+    contributions: z.array(OrganSignalContributionSchema).min(1),
+    corroborating_sources: z.array(z.string().min(1)).min(1),
+    independent_source_count: z.number().int().positive(),
+}));
+/**
+ * `dropped_count` on this section counts *signals* — the fold's input unit —
+ * discarded because their organ did not resolve, because they carried no
+ * locator, or because their organ drew fewer than `min_independent_sources`
+ * distinct sources.
+ */
+export const SafetyCorroborationSchema = withCoverage(z.object({
+    rows: z.array(OrganCorroborationRowSchema),
+    /** Every source id that offered at least one signal on this run. */
+    sources_considered: z.array(z.string().min(1)),
+    /** Distinct sources an organ needed to be reported as corroborated. */
+    min_independent_sources: z.number().int().positive(),
+}));
+// ── §2.6b Claim investigation ───────────────────────────────────────
+/**
+ * What became of a claim once it was argued against.
+ *
+ * Four words, no number. A numeric soundness would need a threshold to mean
+ * anything, and a threshold on this axis deletes real liabilities: the organs
+ * with the thinnest published mechanism are routinely the ones worth the most
+ * attention. `undetermined` is a real outcome — the argument ran and settled
+ * nothing — not a failure to produce one of the other three.
+ */
+export const ClaimVerdictSchema = z.enum(["upheld", "weakened", "overturned", "undetermined"]);
+/** Which stopping condition ended a claim's investigation. */
+export const InvestigationConvergenceSchema = z.enum(["verdict_terminal", "verdict_settled", "round_bound_reached"]);
+/** How the target is proposed to produce the liability in an organ. */
+export const MechanismProposalSchema = withEvidence(z.object({ statement: z.string().min(1) }));
+/**
+ * The case against the claim.
+ *
+ * Its `support` is the counter-evidence, held to the same contract as the claim
+ * it attacks: an objection that cannot cite a record is `unknown` with a reason,
+ * which is what stops "no supporting literature was found" being dressed up as a
+ * finding of its own.
+ */
+export const ClaimCritiqueSchema = withEvidence(z.object({ objection: z.string().min(1) }));
+/**
+ * One organ claim, after interrogation.
+ *
+ * The row's own `support` backs the **verdict** — the assertion the row exists
+ * to make. The mechanism and the critique are nested claims carrying their own
+ * support, because they are separate assertions that can independently fail to
+ * be evidenced; flattening all three onto one `support` would make it
+ * unreadable which of them it was for.
+ *
+ * `critique` is singular: objections are deduplicated by organ, so a later
+ * round's critique replaces the earlier one it was written against, and
+ * `rounds_run` says how many passes it took.
+ */
+export const InvestigatedClaimRowSchema = withEvidence(z.object({
+    organ: OrganSystemSchema,
+    /** Null when no mechanism was proposable from the evidence collected. */
+    mechanism: MechanismProposalSchema.nullable(),
+    /** Null when the critique step produced no objection worth recording. */
+    critique: ClaimCritiqueSchema.nullable(),
+    verdict: ClaimVerdictSchema,
+    rounds_run: z.number().int().positive(),
+    convergence: InvestigationConvergenceSchema,
+}));
+/** Why a candidate organ claim produced no investigated row. */
+export const UninvestigatedClaimReasonSchema = z.enum(["not_corroborated", "exceeded_claim_budget", "investigation_unavailable"]);
+export const UninvestigatedClaimSchema = z.object({
+    organ: OrganSystemSchema,
+    reason: UninvestigatedClaimReasonSchema,
+    detail: z.string().min(1),
 });
+/**
+ * The investigation record.
+ *
+ * `not_investigated` is part of the data rather than a diagnostic beside it: a
+ * run that investigated nothing still has something to say, and it is the only
+ * thing that case has to say. That is why `rows` may be empty on an `available`
+ * section — every candidate that produced no row is named here with its reason.
+ *
+ * The two bounds ride in the section because they are what a reader needs to
+ * know how hard each claim was pushed, and they are configuration, so the value
+ * in force is not derivable from the code a reader happens to be looking at.
+ */
+export const ClaimInvestigationSchema = withCoverage(z.object({
+    rows: z.array(InvestigatedClaimRowSchema),
+    not_investigated: z.array(UninvestigatedClaimSchema),
+    /** Propose → critique → re-verify rounds a single claim was allowed. */
+    round_bound: z.number().int().positive(),
+    /** Corroborated claims this run was allowed to investigate. */
+    claim_budget: z.number().int().positive(),
+}));
 // ── §2.7 Off-tissue risk ────────────────────────────────────────────
-export const OffTissueRowSchema = z.object({
+export const OffTissueRowSchema = withEvidence(z.object({
     tissue: z.string(),
-    organ: z.string(),
+    organ: OrganSystemSchema,
     tpm: z.number(),
-});
+}));
 export const OffTissueRiskSchema = withCoverage(z.object({ rows: z.array(OffTissueRowSchema) }));
 // ── §3.1–3.6 Reference biology — keyed evidence sections ────────────
 export const KeyedEvidenceRowSchema = z.object({
@@ -611,19 +882,40 @@ export const PpiPartnerRowSchema = z.object({
 });
 export const PpiNetworkSchema = withCoverage(z.object({ partners: z.array(PpiPartnerRowSchema) }));
 // ── §3.9 Normal tissue expression ───────────────────────────────────
-export const NormalTissueRowSchema = z.object({
-    tissue: z.string(),
-    organ: z.string().optional(),
-    tpm: z.number(),
-    is_safety_relevant: z.boolean(),
+/** Source and unit travel together, and the normalization note is mandatory. */
+export const NormalTissueExpressionSchema = z
+    .object({
+    source: z.enum(["gtex", "hpa_consensus", "hpa_rna_tissue"]),
+    unit: z.enum(["tpm", "ntpm", "consensus_normalized"]),
+    normalization_notes: z.string().min(1),
+    rows: z.array(z.object({
+        tissue: z.string(),
+        value: z.number(),
+        percentile_rank: z.number().optional(),
+    })),
+})
+    .superRefine((v, ctx) => {
+    if (v.unit === "tpm" && v.source !== "gtex") {
+        ctx.addIssue({
+            code: "custom",
+            message: `unit:"tpm" is only valid with source:"gtex" (got source:"${v.source}")`,
+            path: ["unit"],
+        });
+    }
+    if (v.unit === "ntpm" && !(v.source === "hpa_rna_tissue" || v.source === "hpa_consensus")) {
+        ctx.addIssue({
+            code: "custom",
+            message: `unit:"ntpm" is only valid with HPA sources`,
+            path: ["unit"],
+        });
+    }
 });
-export const NormalTissueExpressionSchema = withCoverage(z.object({ rows: z.array(NormalTissueRowSchema) }));
 // ── §3.10 Preclinical ───────────────────────────────────────────────
 export const KoPhenotypeSchema = z.object({
     marker_symbol: z.string().nullable(),
     viability: z.string().nullable(),
     sex_dimorphism: z.boolean(),
-    organ_systems_with_phenotype: z.array(z.string()),
+    organ_systems_with_phenotype: z.array(OrganSystemSchema),
     top_mp_terms: z.array(z.string()),
     total_phenotype_count: z.number().int().nonnegative(),
     pre_weaning_lethal: z.boolean(),
@@ -713,90 +1005,8 @@ export const AdditionalEvidenceSchema = withCoverage(z.object({
     })),
     score_floor: z.number(),
 }));
-// ── V4 schema refinements ────────────────────────────────────────────
-// Trial outcome with required discriminated effect block
-export const TrialOutcomeRowV4Schema = z.object({
-    nct_id: z.string(),
-    measure: z.string(),
-    outcome_type: z.enum(["primary", "secondary", "other"]),
-    description: z.string().optional(),
-    time_frame: z.string().optional(),
-    effect: z.discriminatedUnion("kind", [
-        z.object({
-            kind: z.literal("quantitative"),
-            value: z.number(),
-            ci_low: z.number().optional(),
-            ci_high: z.number().optional(),
-            units: z.string(),
-        }),
-        z.object({
-            kind: z.literal("qualitative"),
-            direction: z.enum(["increase", "decrease", "no_change"]),
-            magnitude_label: z.string(),
-        }),
-        z.object({
-            kind: z.literal("not_extracted"),
-            reason: z.enum(["ctgov_no_numeric_result", "ctgov_no_result_groups", "parse_failure"]),
-        }),
-    ]),
-});
-// Normal tissue expression with source + unit pair and required normalization_notes
-export const NormalTissueExpressionV4Schema = z
-    .object({
-    source: z.enum(["gtex", "hpa_consensus", "hpa_rna_tissue"]),
-    unit: z.enum(["tpm", "ntpm", "consensus_normalized"]),
-    normalization_notes: z.string().min(1),
-    rows: z.array(z.object({
-        tissue: z.string(),
-        value: z.number(),
-        percentile_rank: z.number().optional(),
-    })),
-})
-    .superRefine((v, ctx) => {
-    if (v.unit === "tpm" && v.source !== "gtex") {
-        ctx.addIssue({
-            code: "custom",
-            message: `unit:"tpm" is only valid with source:"gtex" (got source:"${v.source}")`,
-            path: ["unit"],
-        });
-    }
-    if (v.unit === "ntpm" && !(v.source === "hpa_rna_tissue" || v.source === "hpa_consensus")) {
-        ctx.addIssue({
-            code: "custom",
-            message: `unit:"ntpm" is only valid with HPA sources`,
-            path: ["unit"],
-        });
-    }
-});
-// Tractability with modality enumeration refinement
-const MOLECULE_TYPE_TO_MODALITY = {
-    Protein: ["peptide", "antibody"],
-    "Small molecule": ["small_molecule"],
-    Oligonucleotide: ["oligonucleotide"],
-};
-export const TractabilityV4Schema = withCoverage(z
-    .object({
-    modalities: z.array(TractabilityModalitySchema),
-    preferred_modality: z.string().nullable(),
-    drug_molecule_types_present: z.array(z.string()).default([]),
-})
-    .superRefine((v, ctx) => {
-    for (const moleculeType of v.drug_molecule_types_present) {
-        const allowed = MOLECULE_TYPE_TO_MODALITY[moleculeType];
-        if (!allowed)
-            continue; // unknown molecule_type — skip rather than block
-        const hasMatching = v.modalities.some((m) => allowed.includes(m.modality) && !m.is_inferred_from_family);
-        if (!hasMatching) {
-            ctx.addIssue({
-                code: "custom",
-                message: `drug molecule_type "${moleculeType}" has no matching enumerated modality row`,
-                path: ["modalities"],
-            });
-        }
-    }
-}));
-// Discovery trial row with required relevance_basis discriminator
-export const DiscoveryTrialRowV4Schema = z
+// ── §4.5 Discovery trials ───────────────────────────────────────────
+export const DiscoveryTrialRowSchema = z
     .object({
     nct_id: z.string(),
     title: z.string(),
@@ -816,6 +1026,8 @@ export const DiscoveryTrialRowV4Schema = z
         z.object({ kind: z.literal("condition_match"), matched_term: z.string().optional() }),
         z.object({ kind: z.literal("manual") }),
     ]),
+    attribution: ClinicalTrialAttributionSchema,
+    eligible_for_toxicology_aggregation: z.boolean(),
 })
     .superRefine((v, ctx) => {
     // Low-confidence rows that only match by condition (no drug link) are unreliable;
@@ -828,66 +1040,11 @@ export const DiscoveryTrialRowV4Schema = z
         });
     }
 });
-// ── V4 derived sub-tree ─────────────────────────────────────────────
-// Shape produced by computeDerivedFields(body) and validated at phase-5
-// persist. `.strict()` rejects unknown keys to prevent accidental extras.
-export const DerivedV4Schema = z
-    .object({
-    summary: z.object({
-        has_human_evidence: z.boolean(),
-        has_clinical_evidence: z.boolean(),
-        total_evidence_items: z.number().int().nonnegative(),
-        total_distinct_papers: z.number().int().nonnegative(),
-        total_distinct_clinical_trials: z.number().int().nonnegative(),
-        strongest_strength_label: z.string().nullable(),
-        highest_score: z.number().nullable(),
-        highest_score_is_conflicted: z.boolean(),
-    }),
-    risk_summary: z.object({
-        any_fatal_signal: z.union([z.boolean(), z.literal("unknown")]),
-        highest_risk_organ: z.string().nullable(),
-        off_target_safety_target_hits: z.number().int().nonnegative(),
-        class_liability_count: z.number().int().nonnegative(),
-    }),
-    liability_summary: z.object({
-        counts: z.object({
-            class_liability_count: z.number().int().nonnegative(),
-            safety_target_off_target_count: z.number().int().nonnegative(),
-            off_tissue_risk_organ_count: z.number().int().nonnegative(),
-        }),
-        expression_breadth: z.object({
-            total_assessed_tissues: z.number().int().nonnegative(),
-            high_expression_tissue_count: z.number().int().nonnegative(),
-        }),
-        same_class_drug_count: z.number().int().nonnegative(),
-        safety_data_sources_checked: z.number().int().nonnegative(),
-    }),
-    organ_rollup_completeness: z
-        .object({
-        expected_organs: z.array(z.string()),
-        present_organs: z.array(z.string()),
-        missing_organs: z.array(z.string()),
-    })
-        .refine((v) => v.missing_organs.length === 0, {
-        message: "missing_organs must be empty for the dossier to ship",
-        path: ["missing_organs"],
-    }),
-})
-    .strict();
-// ── §4.5 Discovery trials ───────────────────────────────────────────
-export const DiscoveryTrialRowSchema = ClinicalTrialRowSchema.extend({
-    match_confidence: z.enum(["high", "medium", "low", "off_target"]),
-});
 export const DiscoveryTrialsSchema = withCoverage(z.object({
-    rows: z.array(DiscoveryTrialRowV4Schema),
-    related_target_trials: z
-        .array(ClinicalTrialRowSchema.extend({
-        match_confidence: z.literal("off_target"),
-    }))
-        .optional(),
-    related_receptor: z.string().optional(),
+    rows: z.array(DiscoveryTrialRowSchema),
+    excluded_rows: z.array(DiscoveryTrialRowSchema).default([]),
 }));
-// ── §1.0 Executive recommendation (v2) ──────────────────────────────
+// ── §1.0 Executive recommendation ───────────────────────────────────
 export const ExecutiveRecommendationDataSchema = z.object({
     disposition: z.enum(["pursue", "conditional", "de_prioritize", "insufficient_evidence"]),
     confidence: z.enum(["high", "medium", "low"]),
@@ -901,15 +1058,6 @@ export const ExecutiveRecommendationDataSchema = z.object({
         cited_pmid: z.string().optional(),
     }))
         .optional(),
-    // Deprecated; kept for one release for backward compat with previously
-    // persisted dossiers. New writes use `key_risks_dropped`. v5 will remove.
-    key_risks_downgraded: z
-        .array(z.object({
-        bullet_index: z.number().int().nonnegative(),
-        reason: z.string(),
-        cited_pmid: z.string().optional(),
-    }))
-        .optional(),
     modality_choice: z.object({
         modality: z.string(),
         rationale: z.string(),
@@ -928,7 +1076,7 @@ export const ExecutiveRecommendationDataSchema = z.object({
     }),
 });
 export const ExecutiveRecommendationSchema = withCoverage(ExecutiveRecommendationDataSchema);
-// ── §4.6 Synthesis diagnostics (v2) ─────────────────────────────────
+// ── §4.6 Synthesis diagnostics ──────────────────────────────────────
 export const SynthesisDiagnosticRowSchema = z.object({
     step_id: z.enum(["liability-bullets", "safety-flags-trail", "translational-commentary", "dossier-recommendation"]),
     model_id: z.string(),
@@ -946,7 +1094,7 @@ export const SynthesisDiagnosticRowSchema = z.object({
         "relaxed",
         "skipped",
     ]),
-    final_coverage: z.enum(["available", "queried_no_data"]),
+    final_coverage: CoverageSchema.extract(["available", "queried_no_data"]),
     error_kind: z.enum(["synthesis-too-shallow", "synthesis-unavailable", "voice-violation"]).nullable().optional(),
     error_message: z.string().nullable().optional(),
 });
@@ -1000,11 +1148,18 @@ export const RecommendationAuditSchema = withCoverage(z.object({
     citations_unresolved: z.array(RecommendationAuditEntrySchema),
     non_dossier_citations: z.array(z.object({ token: z.string(), source: z.string() })),
 }));
-// ── Top-level Dossier ───────────────────────────────────────────────
-//
-// `DossierSchema` is a single v3 shape. No backwards-compatibility union.
-// Old v1/v2 persisted rows will not parse; that is intentional.
-const ReferenceBiologyShape = z.object({
+// ── §4.8 Quality gates ──────────────────────────────────────────────
+export const QualityGateStatusSchema = z.object({
+    gate_id: z.string(),
+    status: z.enum(["pass", "warning", "blocked"]),
+    message: z.string(),
+    affected_sections: z.array(z.string()).default([]),
+    unresolved_count: z.number().int().nonnegative().optional(),
+    metadata: z.record(z.string(), z.unknown()).optional(),
+});
+export const QualityGatesSchema = withCoverage(z.object({ rows: z.array(QualityGateStatusSchema) }));
+// ── Composite shapes ────────────────────────────────────────────────
+export const ReferenceBiologyShape = z.object({
     therapeutic_area_associations: DiseaseAssociationsSchema,
     molecular_interactions: MolecularInteractionsSchema,
     biomarker_potential: BiomarkerPotentialSchema,
@@ -1013,7 +1168,7 @@ const ReferenceBiologyShape = z.object({
     combination_evidence: CombinationEvidenceSchema,
     pathway_context: PathwayContextSchema,
     ppi_network: PpiNetworkSchema,
-    normal_tissue_expression: NormalTissueExpressionSchema,
+    normal_tissue_expression: withCoverage(NormalTissueExpressionSchema),
     preclinical: PreclinicalProfileSchema,
     key_papers: KeyPapersSchema,
 });
@@ -1024,300 +1179,88 @@ const AnalyticsShape = z.object({
     additional_evidence: AdditionalEvidenceSchema,
     discovery_trials: DiscoveryTrialsSchema,
     synthesis_diagnostics: SynthesisDiagnosticsSchema,
+    quality_gates: QualityGatesSchema,
     recommendation_audit: RecommendationAuditSchema.optional(),
 });
-const DossierCommonShape = {
-    generated_at: z.string(),
+// ── Derived sub-tree ────────────────────────────────────────────────
+//
+// Shape produced by computeDerivedFields(body) and validated at phase-5
+// persist. `.strict()` rejects unknown keys to prevent accidental extras.
+export const DerivedSchema = z
+    .object({
+    summary: z.object({
+        has_human_evidence: z.boolean(),
+        has_clinical_evidence: z.boolean(),
+        total_evidence_items: z.number().int().nonnegative(),
+        total_distinct_papers: z.number().int().nonnegative(),
+        total_distinct_clinical_trials: z.number().int().nonnegative(),
+        strongest_strength_label: z.string().nullable(),
+        highest_score: z.number().nullable(),
+        highest_score_is_conflicted: z.boolean(),
+    }),
+    risk_summary: z.object({
+        any_fatal_signal: z.union([z.boolean(), z.literal("unknown")]),
+        highest_risk_organ: OrganSystemSchema.nullable(),
+        off_target_safety_target_hits: z.number().int().nonnegative(),
+        class_liability_count: z.number().int().nonnegative(),
+    }),
+    liability_summary: z.object({
+        counts: z.object({
+            class_liability_count: z.number().int().nonnegative(),
+            safety_target_off_target_count: z.number().int().nonnegative(),
+            off_tissue_risk_organ_count: z.number().int().nonnegative(),
+        }),
+        expression_breadth: z.object({
+            total_assessed_tissues: z.number().int().nonnegative(),
+            high_expression_tissue_count: z.number().int().nonnegative(),
+        }),
+        same_class_drug_count: z.number().int().nonnegative(),
+        safety_data_sources_checked: z.number().int().nonnegative(),
+    }),
+    organ_rollup_completeness: z
+        .object({
+        expected_organs: z.array(OrganSystemSchema),
+        present_organs: z.array(OrganSystemSchema),
+        missing_organs: z.array(OrganSystemSchema),
+    })
+        .refine((v) => v.missing_organs.length === 0, {
+        message: "missing_organs must be empty for the dossier to ship",
+        path: ["missing_organs"],
+    }),
+})
+    .strict();
+// ── Top-level Dossier ───────────────────────────────────────────────
+export const DossierSchema = z.object({
     entity: EntitySchema,
-    summary: SummarySchema,
+    generated_at: z.string(),
     liability_summary: LiabilitySummarySchema,
     tractability: TractabilitySchema,
     indications: IndicationsSchema,
     drug_interactions: DrugInteractionsSchema,
     clinical_development: ClinicalDevelopmentSchema,
     safety_profile: SafetyProfileSchema,
+    safety_corroboration: SafetyCorroborationSchema,
+    claim_investigation: ClaimInvestigationSchema,
     off_tissue_risk: OffTissueRiskSchema,
-    off_target_panel: withCoverage(z.object({ rows: z.array(OffTargetRowSchema) })),
+    off_target_panel: withCoverage(OffTargetPanelSchema),
     reference_biology: ReferenceBiologyShape,
-};
-export const DossierSchema = z.object({
-    schema_version: z.literal("3"),
-    ...DossierCommonShape,
-    analytics: AnalyticsShape,
-    executive_recommendation: ExecutiveRecommendationSchema,
-});
-// ── V4 section wrappers ─────────────────────────────────────────────
-// OrganRiskRow V4 — extends v3 with a richer organ enum
-export const OrganRiskRowV4Schema = OrganRiskRowSchema.extend({
-    organ: z.enum([
-        "cardiac",
-        "hepatic",
-        "cns",
-        "renal",
-        "gi",
-        "pancreas",
-        "endocrine_thyroid",
-        "metabolic",
-        "hematologic",
-        "immune",
-        "respiratory",
-        "reproductive",
-        "dermatologic",
-        "musculoskeletal",
-        "oncology",
-    ]),
-});
-// Liability bullets removed post-synthesis because they cited a PMID the
-// direction-of-effect auditor found contradicts the bullet's framing.
-// Mirrors `executive_recommendation.key_risks_dropped` — the dropped
-// content is preserved with the reason for an audit trail.
-export const LiabilityBulletDroppedSchema = LiabilityBulletSchema.extend({
-    reason: z.string(),
-    cited_pmid: z.string().optional(),
-});
-// LiabilitySummaryV4Schema — strict() rejects derived counter keys that moved to derived sub-tree
-export const LiabilitySummaryV4Schema = z
-    .object({
-    liability_bullets: z.array(LiabilityBulletSchema),
-    liability_bullets_dropped: z.array(LiabilityBulletDroppedSchema).optional(),
-    modality_recommendation: z.string().nullable(),
-    same_class_drug_count: z.number().int().nonnegative(),
-    inferred_therapeutic_area: z.string().nullable(),
-    no_liabilities_disclosure: z.string().optional(),
-})
-    .strict();
-// ClinicalDevelopmentV4Schema — swaps in V4 outcome and failed-trial rows
-export const ClinicalDevelopmentV4Schema = z.object({
-    trials: withCoverage(z.object({
-        rows: z.array(ClinicalTrialRowSchema),
-        selection_criteria: z
-            .object({
-            derived_from: z.string().optional(),
-            min_confidence: z.string().optional(),
-            excluded_off_target_count: z.number().int().nonnegative().optional(),
-        })
-            .optional(),
-    })),
-    outcomes: withCoverage(z.object({ rows: z.array(TrialOutcomeRowV4Schema) })),
-    failed_trials: withCoverage(z.object({ rows: z.array(FailedTrialRowV4Schema) })),
-    benchmarks: ClinicalBenchmarksSchema,
-});
-const RegulatoryActionRowSchema = z.object({
-    drug_chembl_id: z.string(),
-    drug_name: z.string(),
-    agency: z.enum(["FDA", "EMA", "MHRA", "PMDA", "Health Canada", "TGA"]),
-    action_kind: z.enum(["referral", "withdrawal", "indication_restriction", "REMS", "black_box", "DHCP", "safety_communication", "label_warning"]),
-    action_date: z.string(),
-    finding: z.string(),
-    source_url: z.string().url().optional(),
-    source_kind: z.enum(["label_warning", "referral", "rems", "withdrawal", "boxed_warning", "safety_communication"]).optional(),
-    application_number: z.string().optional(),
-    label_section: z.string().optional(),
-    source_date: z.string().optional(),
-    evidence: EvidenceList,
-});
-// SafetyProfileV4Schema — swaps in V4 FAERS / off-target / failed-trial / OrganRiskRow
-export const SafetyProfileV4Schema = z.object({
-    organ_rollup: withCoverage(z.object({ rows: z.array(OrganRiskRowV4Schema) })),
-    faers: withCoverage(FaersSummaryV4Schema),
-    trial_aes: withCoverage(TrialAesSchema),
-    off_target_panel: withCoverage(OffTargetPanelV4Schema),
-    failed_trials_safety_lens: withCoverage(z.object({ rows: z.array(FailedTrialRowV4Schema) })),
-    class_precedent: withCoverage(ClassPrecedentSchema),
-    target_organ_liabilities: z.array(SafetyFlagSchema),
-    regulatory_actions: withCoverage(z.object({ rows: z.array(RegulatoryActionRowSchema) })).optional(),
-});
-// ReferenceBiologyV4Shape — swaps normal_tissue_expression for V4 version
-export const ReferenceBiologyV4Shape = z.object({
-    ...ReferenceBiologyShape.shape,
-    normal_tissue_expression: withCoverage(NormalTissueExpressionV4Schema),
-});
-// ── V4 top-level DossierV4Schema ────────────────────────────────────
-export const DossierV4Schema = z.object({
-    schema_version: z.literal("4"),
-    entity: EntitySchema,
-    generated_at: z.string(),
-    liability_summary: LiabilitySummaryV4Schema,
-    tractability: TractabilityV4Schema,
-    indications: IndicationsSchema,
-    drug_interactions: DrugInteractionsSchema,
-    clinical_development: ClinicalDevelopmentV4Schema,
-    safety_profile: SafetyProfileV4Schema,
-    off_tissue_risk: OffTissueRiskSchema,
-    off_target_panel: withCoverage(OffTargetPanelV4Schema),
-    reference_biology: ReferenceBiologyV4Shape,
-    analytics: AnalyticsShape,
-    executive_recommendation: ExecutiveRecommendationSchema,
-    derived: DerivedV4Schema,
-});
-// Body-only v4 shape: everything the Phase-4 assembler is responsible for
-// building, WITHOUT the derived sub-tree. Phase-5 persist computes derived
-// from the stamped body and validates the complete dossier against
-// DossierV4Schema. Keeping Phase4OutputSchema loose prevents an assembler
-// organ-rollup miss from surfacing as a generic workflow-error instead
-// of the explicit derived-invariant-violation path in phase5-persist.
-export const DossierV4BodySchema = DossierV4Schema.omit({ derived: true });
-// ── V5 evidence attribution and quality gates ────────────────────────
-export const ClinicalTrialAttributionBasisV5Schema = z.object({
-    kind: z.enum([
-        "mechanism_target_match",
-        "known_class_drug",
-        "therapeutic_program_match",
-        "related_family_target",
-        "biomarker_endpoint",
-        "condition_only",
-        "text_match",
-        "manual",
-    ]),
-    source: z.string(),
-    excerpt: z.string().optional(),
-    metadata: z.record(z.string(), z.unknown()).optional(),
-});
-export const ResolvedTrialInterventionV5Schema = z.object({
-    name: z.string(),
-    intervention_type: z.string().nullable().optional(),
-    chembl_id: z.string().nullable().optional(),
-    therapeutic_program_id: z.string().nullable().optional(),
-    target_uniprots: z.array(z.string()).default([]),
-    resolver_source: z.string(),
-});
-export const ClinicalTrialAttributionV5Schema = z
-    .object({
-    relationship: z.enum([
-        "direct_modulator",
-        "class_modulator",
-        "related_family_target",
-        "target_biomarker",
-        "pathway_biomarker",
-        "condition_only",
-        "unrelated",
-        "unknown",
-    ]),
-    evidence_role: z.enum(["supports_target", "contextual", "excluded"]),
-    basis: z.array(ClinicalTrialAttributionBasisV5Schema).min(1),
-    resolved_interventions: z.array(ResolvedTrialInterventionV5Schema).default([]),
-    exclusion_reason: z.string().optional(),
-})
-    .superRefine((v, ctx) => {
-    if (v.evidence_role !== "supports_target" && !v.exclusion_reason) {
-        ctx.addIssue({
-            code: "custom",
-            message: "contextual/excluded trial evidence requires exclusion_reason",
-            path: ["exclusion_reason"],
-        });
-    }
-    if (v.evidence_role === "supports_target" && v.relationship !== "direct_modulator" && v.relationship !== "class_modulator") {
-        ctx.addIssue({
-            code: "custom",
-            message: "supports_target evidence must be direct_modulator or class_modulator",
-            path: ["relationship"],
-        });
-    }
-});
-export const ClinicalTrialRowV5Schema = ClinicalTrialRowSchema.extend({
-    attribution: ClinicalTrialAttributionV5Schema,
-    eligible_for_toxicology_aggregation: z.boolean(),
-});
-export const FailedTrialRowV5Schema = FailedTrialRowV4Schema.extend({
-    attribution: ClinicalTrialAttributionV5Schema,
-    eligible_for_toxicology_aggregation: z.boolean(),
-});
-export const TrialOutcomeRowV5Schema = TrialOutcomeRowV4Schema.extend({
-    attribution: ClinicalTrialAttributionV5Schema.optional(),
-    eligible_for_toxicology_aggregation: z.boolean().default(false),
-});
-export const ClinicalDevelopmentV5Schema = z.object({
-    trials: withCoverage(z.object({
-        rows: z.array(ClinicalTrialRowV5Schema),
-        excluded_rows: z.array(ClinicalTrialRowV5Schema).default([]),
-        selection_criteria: z
-            .object({
-            derived_from: z.string().optional(),
-            min_confidence: z.string().optional(),
-            excluded_off_target_count: z.number().int().nonnegative().optional(),
-        })
-            .optional(),
-    })),
-    outcomes: withCoverage(z.object({ rows: z.array(TrialOutcomeRowV5Schema) })),
-    failed_trials: withCoverage(z.object({
-        rows: z.array(FailedTrialRowV5Schema),
-        excluded_rows: z.array(FailedTrialRowV5Schema).default([]),
-    })),
-    benchmarks: ClinicalBenchmarksSchema,
-});
-export const RegulatoryActionRowV5Schema = RegulatoryActionRowSchema.extend({
-    action_kind: z.enum(["referral", "withdrawal", "indication_restriction", "REMS", "black_box", "DHCP", "safety_communication", "label_warning"]),
-    source_kind: z.enum(["label_warning", "referral", "rems", "withdrawal", "boxed_warning", "safety_communication"]),
-    application_number: z.string().optional(),
-    label_section: z.string().optional(),
-    source_date: z.string().optional(),
-}).superRefine((v, ctx) => {
-    if (v.source_kind === "label_warning" && v.action_kind === "safety_communication") {
-        ctx.addIssue({
-            code: "custom",
-            message: "FDA label warnings must not be represented as action_kind:safety_communication",
-            path: ["action_kind"],
-        });
-    }
-});
-export const SafetyProfileV5Schema = SafetyProfileV4Schema.extend({
-    regulatory_actions: withCoverage(z.object({ rows: z.array(RegulatoryActionRowV5Schema) })).optional(),
-});
-export const QualityGateStatusSchema = z.object({
-    gate_id: z.string(),
-    status: z.enum(["pass", "warning", "blocked"]),
-    message: z.string(),
-    affected_sections: z.array(z.string()).default([]),
-    unresolved_count: z.number().int().nonnegative().optional(),
-    metadata: z.record(z.string(), z.unknown()).optional(),
-});
-export const QualityGatesSchema = withCoverage(z.object({ rows: z.array(QualityGateStatusSchema) }));
-export const DiscoveryTrialRowV5Schema = DiscoveryTrialRowV4Schema.extend({
-    attribution: ClinicalTrialAttributionV5Schema,
-    eligible_for_toxicology_aggregation: z.boolean(),
-});
-export const DiscoveryTrialsV5Schema = withCoverage(z.object({
-    rows: z.array(DiscoveryTrialRowV5Schema),
-    excluded_rows: z.array(DiscoveryTrialRowV5Schema).default([]),
-}));
-const AnalyticsV5Shape = AnalyticsShape.extend({
-    discovery_trials: DiscoveryTrialsV5Schema,
-    quality_gates: QualityGatesSchema,
-    recommendation_audit: RecommendationAuditSchema.optional(),
-});
-export const DossierV5Schema = z.object({
-    schema_version: z.literal("5"),
-    entity: EntitySchema,
-    generated_at: z.string(),
-    liability_summary: LiabilitySummaryV4Schema,
-    tractability: TractabilityV4Schema,
-    indications: IndicationsSchema,
-    drug_interactions: DrugInteractionsSchema,
-    clinical_development: ClinicalDevelopmentV5Schema,
-    safety_profile: SafetyProfileV5Schema,
-    off_tissue_risk: OffTissueRiskSchema,
-    off_target_panel: withCoverage(OffTargetPanelV4Schema),
-    reference_biology: ReferenceBiologyV4Shape,
-    analytics: AnalyticsV5Shape,
-    executive_recommendation: ExecutiveRecommendationSchema,
-    derived: DerivedV4Schema,
-});
-export const DossierV5BodySchema = DossierV5Schema.omit({ derived: true });
-// Retain the v3 export under a legacy alias for any code that still reads
-// persisted v3 dossiers (phase4-assemble, tests, integration).
-export const DossierSchemaV3Legacy = z.object({
-    schema_version: z.literal("3"),
-    ...DossierCommonShape,
     analytics: AnalyticsShape,
     executive_recommendation: ExecutiveRecommendationSchema,
+    derived: DerivedSchema,
 });
+/**
+ * Everything the Phase-4 assembler is responsible for building, without the
+ * derived sub-tree. Phase-5 persist computes derived from the stamped body
+ * and validates the complete dossier against `DossierSchema`.
+ */
+export const DossierBodySchema = DossierSchema.omit({ derived: true });
 // ── Type guards ─────────────────────────────────────────────────────
 export function isDossier(value) {
-    return DossierV5Schema.safeParse(value).success;
+    return DossierSchema.safeParse(value).success;
 }
 /** JSON Schema export for runtime validation in non-TS environments. */
 export function dossierJsonSchema() {
-    return z.toJSONSchema(DossierV5Schema);
+    return z.toJSONSchema(DossierSchema);
 }
 // ── Progress events ─────────────────────────────────────────────────
 export const TargetAssessmentPhaseSchema = z.enum([