FEP dossier — APA 2024 + PORT + RAISE-ETP + NAVIGATE + NICE 2024 + Leucht 2013; depth-pass-1 reconciled 2026-05-14. Step 1: aripiprazole 5–10 mg or risperidone 1–2 mg (lower doses than chronic schizophrenia per PORT Buchanan Schizophr Bull 2010). Step 2: switch to olanzapine 5–10 mg, quetiapine 50–400 mg, or ziprasidone 20–80 mg BID (Leucht Lancet 2013). Step 3: clozapine 12.5 mg → 300–600 mg for treatment-resistant (PORT Buchanan Schizophr Bull 2010; NICE 2024; APA 2024). LAI early: aripiprazole Maintena or paliperidone palmitate if adherence concern (Subotnik AJP 2015; APA 2024). CSC model: EPPIC (McGorry) backbone + RAISE-ETP (Kane AJP 2016 PMID 26844794) + NAVIGATE (Robinson NEJM 2015 PMID 26482160) + OPUS (Petersen 2008) — ≥ 2 years multidisciplinary team-based care; consider ≥ 5 years for select patients per OPUS long-term follow-up; UK NICE recommends ≥ 3 years EIS. DUP minimisation is the single highest-leverage variable in FEP (Marshall Arch Gen Psychiatry 2005); target DUP < 3 months; activate CSC referral SAME-DAY at first FEP presentation, not at follow-up. Mandatory organic workup: UDS, TSH, B12, HIV, RPR, MRI brain, ECG, metabolic baseline (APA 2024; NICE 2024). Severity triggers: acute agitation, SI, command hallucinations, catatonia, NMS, metabolic syndrome, autoimmune_encephalitis_features (NEW), substance_induced_psychosis_not_resolving (NEW), treatment_resistant_psychosis (NEW). Setting playbooks: ED (acute agitation + organic workup), inpatient (stabilisation + CSC transition), outpatient (CSC programme). Inpatient med-psych for organic workup encoded indirectly via autoimmune_encephalitis_features severity trigger + neurology routing — first-class DossierSetting value schema-blocked. Anti-pattern warnings baked in: do NOT delay clozapine after TRS criteria met (APA 2020 / NICE 2024 / PORT Buchanan 2010 PMID 20513679); do NOT use haloperidol if avoidable in FEP (EPS + TD risk in AP-naïve patients per Leucht 2013); do NOT combine olanzapine IM + benzodiazepine (FDA 2004 respiratory depression); do NOT defer organic workup when autoimmune-encephalitis features present (Dalmau Lancet Neurol 2011 PMID 21163445; Graus 2016 criteria); do NOT initiate AP during active catatonia (precipitates NMS — treat catatonia first with lorazepam challenge + ECT if refractory); do NOT discharge FEP without CSC referral activated (DUP minimisation); do NOT skip metabolic baseline (APA/ADA 2004 consensus); do NOT use SSRI monotherapy in unrecognised bipolar with first psychotic episode (mood-switch risk). PRODUCTION blockers: (1) RxCUIs not yet fully validated via npm run research:rxnav (aripiprazole, risperidone, olanzapine, quetiapine, ziprasidone, paliperidone palmitate LAI, aripiprazole Maintena LAI, haloperidol, lorazepam, clozapine); (2) dedicated psych.first-episode-psychosis.core.v1.ts seed manifest not yet authored (manifest field repointed to psych.depression.core.v1.ts per parallel precedent in psych.suicidality + psych.alcohol_withdrawal); (3) FEP-specific calc.* entries (calc.panss, calc.bprs, calc.cssrs, calc.bush_francis, calc.aims, calc.gaf) not yet in clinical-tools-registry — referenced inline in setting_playbooks; (4) FEP-specific panel entries (panel.uds, panel.hiv, panel.rpr, panel.bmp) not registered — UDS / HIV / RPR pursued via inline lab orders, panel.bmp replaced by registered panel.renal; (5) no targeted dossier test file (relies on dossier-contract.test.ts); (6) several PMIDs flagged NEEDS_SOURCE_REVIEW pending Stage-A PubMed verification (Marshall 2005, Petersen 2008 OPUS, Graus 2016, Titulaer 2013, Kane 1988, Sullivan 2012, APA/ADA 2004 consensus, Yatham CANMAT 2018, McGorry EPPIC). Deepened 2026-05-14 (shard-5-obped-id depth-pass-1): authored first co-located design brief at _briefs/psych.first-episode-psychosis.core.v1.md (closes broken design_brief pointer FAIL) + co-located _research-bundles/psych.first-episode-psychosis.core.v1.md (research bundle anchoring EPPIC / RAISE-ETP / NAVIGATE / OPUS / PORT / APA 2020 / NICE 2024 / Leucht 2013 / Dalmau 2011 / Graus 2016 / Titulaer 2013 / Marshall 2005 with verification-rule discipline). Repointed broken manifest pointer prisma/seed/manifests/psych.first-episode-psychosis.core.v1.ts → prisma/seed/manifests/psych.depression.core.v1.ts per parallel precedent (suicidality + alcohol_withdrawal). Repointed 5 unresolved workup IDs to canonical registered umbrellas: workup.suicide_risk_fep → workup.suicide_risk (registered); workup.catatonia_screen → workup.delirium (registered DELIRIUMS umbrella); workup.nms_screen → workup.delirium (NMS rigidity + AMS overlaps delirium umbrella; Levenson criteria inline); workup.autoimmune_encephalitis → workup.anti_nmdar_encephalitis (registered umbrella); workup.organic_psychosis_screen → workup.delirium (organic differential overlaps DELIRIUMS bundle; specific FEP panel inline). Removed 6 unregistered calculator IDs (calc.panss, calc.bprs, calc.cssrs, calc.bush_francis, calc.aims, calc.gaf) — PANSS / BPRS / C-SSRS / Bush-Francis / AIMS / GAF referenced inline in setting_playbooks / severity_triggers / monitoring; retained registered psych calcs (calc.phq9, calc.gad7, calc.audit_c, calc.cows) per suicidality / alcohol_withdrawal precedent. Removed 4 unregistered panel IDs (panel.bmp, panel.uds, panel.hiv, panel.rpr) — UDS / HIV / RPR pursued via inline lab orders; panel.bmp replaced by registered panel.renal (suicidality precedent). Deepened 2026-05-14: added 3 severity_triggers (9 total): autoimmune_encephalitis_features (severe — subacute behavioural change + seizures + dyskinesia + autonomic + cognitive decline + CSF lymphocytic / EEG / MRI abnormality + young / female — Graus 2016 criteria + Dalmau 2011 PMID 21163445; routes to workup.anti_nmdar_encephalitis + neurology + paraneoplastic workup + first-line steroid 1 g IV methylprednisolone + IVIG 0.4 g/kg OR PLEX + tumour resection if paraneoplastic + second-line rituximab + cyclophosphamide for refractory; antipsychotic monotherapy is INSUFFICIENT and may mask the diagnosis); substance_induced_psychosis_not_resolving (severe — DSM-5-TR 2022 1-month rule; psychosis persisting > 1 month after substance cessation → reclassify as primary psychotic disorder, initiate AP at low dose, activate CSC enrollment, continued substance use treatment, DUP clock starts at original symptom onset); treatment_resistant_psychosis (severe — failure of 2 adequate AP trials × adequate dose × 4-6 weeks each = TRS per APA 2020 / NICE 2024 / PORT Buchanan 2010 PMID 20513679 / Kane 1988 / Meltzer 2003 InterSePT; clozapine 12.5 mg PO daily slow titration → therapeutic level 350-600 ng/mL with REMS ANC monitoring weekly × 6 mo / biweekly × 6 mo / monthly thereafter + CK myocarditis screen first 8 weeks + metabolic baseline q 3 mo + seizure-threshold-aware > 600 mg/day + hypersalivation / orthostasis / constipation / ileus screen; DO NOT delay clozapine after TRS criteria met). Deepened 2026-05-14: appended Dalmau Lancet Neurol 2011 PMID 21163445 (anti-NMDA receptor encephalitis canonical case-series + biological characterisation) to evidence.pmids (8 → 9 verified anchors); bumped last_reconciled 2026-05-13 → 2026-05-14. Marshall 2005 (DUP), Petersen 2008 OPUS, Graus 2016, Titulaer 2013, Kane 1988, Sullivan 2012, APA/ADA 2004, Yatham CANMAT 2018, and McGorry EPPIC referenced in research bundle as authoritative anchors but PMIDs not added per verification rule — deferred to depth-pass-2. Phenotype matrix (primary-psychiatric-category × secondary-cause-organic × onset-tempo × first-episode-vs-relapse × catatonia × NMS-history × DUP-tier × insight × host) is encoded indirectly via severity_triggers + setting_playbooks + sibling_differentiation. First-class TS field for phenotype matrix is schema-blocked — see id.sepsis.core.v1 brief Schema-blocked queue. Bayesian linkage (anti-NMDAR CSF IgG against GluN1 sensitivity ≈ 100% / specificity ≈ 100% — LR+ approaching ∞ in clinical context; MRI brain LR+ very high for organic cause if positive; UDS LR+ ≈ 10 for substance-induced differential; PANSS ≥ 20% reduction by week 2 LR+ moderate for longer-term response per Leucht 2013 PMID 23498057; first-degree relative with schizophrenia LR+ ≈ 10; first-degree relative with bipolar LR+ ≈ 3-4 — shifts toward mood-disorder-with-psychotic-features differential; T_treat = at FEP confirmation, initiate AP at low dose + activate CSC simultaneously, do NOT wait for full diagnostic clarity in first 6 weeks per RAISE-ETP; T_test = brief psychotic disorder < 1 month + clear stressor + intact insight + safety preserved + supports engaged → observation candidacy WITH short-course low-dose AP; T_clozapine = failure of 2 adequate AP trials × adequate dose × 4-6 weeks each, do NOT delay; cross-dossier routing to psych.bipolar-disorder.core.v1 / psych.depression.core.v1 / psych.suicidality.ed.core.v1 / future neuro.autoimmune-encephalitis.core.v1 via workup.anti_nmdar_encephalitis umbrella / workup.delirium for geriatric organic differential / psych.alcohol_withdrawal.core.v1 / psych.opioid_use_disorder.core.v1) is documented in the co-located _research-bundles/psych.first-episode-psychosis.core.v1.md.