feat: v2.0.0 — 20-module pipeline, Phase 5 complete, 180 tests

fix: methodology summary fix — use r.summary instead of llm_assessment

feat: add PDF upload endpoint with 14-module analysis and security hardening

feat: section-aware text routing — each module gets relevant paper sections

feat: add LLM-Generated Paper Detector - Phase 2 COMPLETE

feat: add Citation Cartel Detector - network manipulation detector

feat: add Retraction Watch Checker - live fraud database matching

feat: add Effect Size Validator - power analysis + inflated effect detector

fix: add smart text truncation 8000 chars - long paper support

feat: add P-Curve Analyzer - publication bias detector

feat: add Granularity Analyzer - Benford Law + digit preference detector

feat: add SPRITE Test module - impossible distribution detector

fix: novelty 500 error - getattr risk_score and flags fallback

feat: add GRIM Test module - mathematically impossible means detector

fix: initialize _repro_engine and _novelty_engine in routes.py

fix: CORS in create_app factory, clean main.py

feat: novelty scorer — 8/8 tests, all 6 modules complete, 45 total tests

feat: reproducibility scanner — 10/10 tests, code/data/ethics detection, API endpoint

fix: status endpoint updated — 4 modules marked ready

feat: citation integrity analyzer — self-citation, unsupported claims, density — 27 total tests

feat: methodology checker live — 3 flags detected, causation/control/timeframe — 20 total tests passing

feat: figure forensics API endpoint — PDF upload, forensic analysis live

feat: live stat audit API working — detects p-hacking, small samples, critical risk scoring

feat: statistical audit engine with 7 tests — all passing

feat: SciPeerAI project foundation initialized