diff --git a/.github/workflows/ingest-smoke.yml b/.github/workflows/ingest-smoke.yml
new file mode 100644
index 0000000..c235068
--- /dev/null
+++ b/.github/workflows/ingest-smoke.yml
@@ -0,0 +1,75 @@
+name: Ingest smoke (dev S3)
+
+# Separate from PR CI: proves GitHub OIDC → AWS dev bucket access.
+# Requires GitHub Environment "dev" with repository variables (see docs/mvp/guides/cloud-foundation.md).
+
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: "0 6 * * 1"
+
+permissions:
+  id-token: write
+  contents: read
+
+jobs:
+  ingest-smoke:
+    runs-on: ubuntu-latest
+    environment: dev
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Configure AWS credentials (OIDC)
+        uses: aws-actions/configure-aws-credentials@v4
+        with:
+          role-to-assume: ${{ vars.AWS_ROLE_ARN }}
+          aws-region: ${{ vars.AWS_REGION }}
+
+      - name: Write smoke marker to dev lake raw zone
+        env:
+          DEV_LAKE_BUCKET: ${{ vars.DEV_LAKE_BUCKET }}
+          DEV_LAKE_PREFIX: ${{ vars.DEV_LAKE_PREFIX }}
+        run: |
+          set -euo pipefail
+          if [ -z "${DEV_LAKE_BUCKET:-}" ]; then
+            echo "Missing DEV_LAKE_BUCKET. See docs/mvp/guides/cloud-foundation.md"
+            exit 1
+          fi
+          prefix="${DEV_LAKE_PREFIX:-}"
+          key="${prefix}raw/_smoke/ingest-smoke.txt"
+          echo "smartwealthai ingest-smoke $(date -u +%Y-%m-%dT%H:%M:%SZ)" > smoke.txt
+          aws s3 cp smoke.txt "s3://${DEV_LAKE_BUCKET}/${key}"
+
+      - name: Validate smoke object
+        env:
+          DEV_LAKE_BUCKET: ${{ vars.DEV_LAKE_BUCKET }}
+          DEV_LAKE_PREFIX: ${{ vars.DEV_LAKE_PREFIX }}
+        run: |
+          set -euo pipefail
+          prefix="${DEV_LAKE_PREFIX:-}"
+          key="${prefix}raw/_smoke/ingest-smoke.txt"
+          aws s3api head-object --bucket "${DEV_LAKE_BUCKET}" --key "${key}"
+          aws s3 cp "s3://${DEV_LAKE_BUCKET}/${key}" -
+          echo
+          echo "ingest-smoke OK: s3://${DEV_LAKE_BUCKET}/${key}"
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install Poetry
+        uses: snok/install-poetry@76e04a911780d5b312d89783f7b1cd627778900a
+        with:
+          virtualenvs-in-project: true
+
+      - name: Validate lake root URI parsing (no live S3 I/O)
+        run: |
+          poetry install --with dev
+          poetry run python -c "
+          from smartwealthai.lake_root import resolve_lake_root
+          root = resolve_lake_root('s3://${{ vars.DEV_LAKE_BUCKET }}/${{ vars.DEV_LAKE_PREFIX }}')
+          assert root.backend == 's3'
+          print('lake root:', root.uri, 'bucket=', root.s3_bucket)
+          "
diff --git a/CONTEXT.md b/CONTEXT.md
index abc9705..dcd9c09 100644
--- a/CONTEXT.md
+++ b/CONTEXT.md
@@ -77,12 +77,40 @@ _Avoid_: dividend yield, earnings/price without EV
 _Avoid_: market cap alone as “value”
 
 **Combined rank**:
-Sum of ROC rank and EY rank; lower is better. Used for portfolio selection and sell-watch opportunity cost.
-_Avoid_: average of ranks, z-score blend (not MVP)
+Sum of ROC rank and EY rank; lower is better. **Benchmark only** (Magic Formula replica) — not production portfolio selection after Phase 2. Sell-watch opportunity-cost triggers on MF path are superseded by QV triggers in production.
+_Avoid_: production ranking term post–Phase 2; average of ranks, z-score blend
 
 **Magic Formula replica**:
-Canonical benchmark portfolio using the same ROC, EY, combined rank, universe, and annual rebalance as production. Used to gate backtest pass vs strategy Sharpe.
-_Avoid_: live Greenblatt fund, generic “value factor”
+Canonical **benchmark** portfolio using ROC, EY, combined rank, universe, and annual rebalance. Used to gate backtest pass vs strategy Sharpe. **Not** the Phase 2+ production scoring path.
+_Avoid_: live Greenblatt fund, generic “value factor”, conflating with Quantitative Value production
+
+**Quantitative Value (QV) funnel**:
+Phase 2+ **production** scoring: universe → forensic hard exclusion (incl. Beneish bottom-5%) → EBIT/TEV value decile → FS-Score quality screen → ~50-name equal-weight model portfolio. Spec: `docs/mvp/features/quantitative-value.md`.
+_Avoid_: Magic Formula path, ROC+EY combined rank for production
+
+**Quality (production)**:
+**FS-Score** composite (0–10, Gray/Carlisle variant) on the EBIT/TEV value pool — not ROC rank alone.
+_Avoid_: ROC rank as production quality after Phase 2; ESG or subjective moat
+
+**Cheap (production)**:
+Membership in the **EBIT/TEV value pool** (top decile among forensic survivors) — not EY rank alone.
+_Avoid_: EY rank as production cheapness after Phase 2; low P/E without EV
+
+**EBIT/TEV**:
+`EBIT / Enterprise Value`; value-screen metric for QV production. Same EV definition as **Earnings yield**; ranked within forensic survivors to form the value pool. Spec: `docs/mvp/features/quantitative-value.md`.
+_Avoid_: MF EY rank, market cap alone
+
+**FS-Score**:
+Ten binary financial-strength components (profitability, stability, recent operational improvements) summed to 0–10. Production quality factor after Phase 2. `formula_version` on every scored row.
+_Avoid_: Piotroski F-Score (different formula), ROC as production quality
+
+**QV funnel rank**:
+Order within the value pool after the FS-Score quality screen; determines portfolio membership. Supersedes **combined rank** for production.
+_Avoid_: combined rank, ROC rank + EY rank for production selection
+
+**Forensic evaluator**:
+Hard `exclude` / `pass` before value or quality scoring; distress and fraud rules from permanent-loss filter plus Beneish M-Score bottom-5% gate. Every exclusion carries `rule_id`, `rule_version`, `triggered_value`, `threshold`, `explanation`.
+_Avoid_: soft penalty, scoring before forensics
 
 **Cross-sectional rank**:
 Rank across all passing companies on one run date. Not comparable across dates without re-running the pipeline.
@@ -93,7 +121,7 @@ Version id for ROC, EY, or filter rules so runs and backtests stay reproducible.
 _Avoid_: “latest formula”, implicit default
 
 **Model portfolio**:
-Target long-only holdings from the pipeline; **June 30 demo:** top 30 names by combined rank, equal-weight only, market-cap tie-break on ranks. No watchlist in demo slice. Paper-traded in full MVP (phase 2).
+Target long-only holdings from the pipeline. **June 30 demo:** top 30 by MF combined rank, equal-weight, market-cap tie-break. **Phase 2+ production (QV):** ~50 names by FS-Score within the EBIT/TEV value pool, equal-weight, market-cap tie-break (configurable cap). Paper-traded in full MVP (phase 2).
 _Avoid_: personal portfolio, watchlist (demo slice)
 
 **Watchlist**:
@@ -130,7 +158,9 @@ _Avoid_: ad-hoc snapshot without run id
 
 ## Relationships
 
-- A **run date** drives **universe** → **permanent loss filter** → **ROC** and **EY** ranks → **combined rank** → **model portfolio**
+- **Demo:** **run date** → **universe** → **ROC** and **EY** ranks → **combined rank** → **model portfolio** (top 30)
+- **Phase 2+ production:** **run date** → **universe** → **forensic evaluator** → **EBIT/TEV value pool** → **FS-Score** → **QV funnel rank** → **model portfolio** (~50)
+- **Benchmark (all phases):** MF replica path (ROC + EY + combined rank) for backtest Sharpe gate — parallel to production, not mixed into QV funnel
 - **As-of date** tags each fundamental row; PIT queries filter `as_of_date <= run_date`
 - **Watchlist** superset of names that may enter the **model portfolio** on rebalance
 - **Magic Formula replica** is the strategy’s primary benchmark comparator for Sharpe pass/fail
@@ -142,11 +172,12 @@ Resolved scope cuts (see ADRs and [`docs/mvp/demo-slice.md`](docs/mvp/demo-slice
 
 - **June 30 demo MVP:** SimFin bulk US → raw → normalizer → **universe (US market)** → ROC/EY → combined rank → top-30 EW model portfolio → Streamlit dashboard. No permanent loss filter, backtest, sell-watch, or paper trading in this slice.
 - SEC ETL spike (`sec_client`, `edgartools_client`, `download-fundamentals`) is **frozen** in repo for phase 2; demo pipeline uses SimFin bulk for fundamentals and run-date prices (`shareprices/latest`).
-- **Phase 2 (Quantitative Value):** will need multi-period fundamentals (not only TTM snapshots)—lake design should not block adding annual/quarterly income history later.
+- **Phase 2 (Quantitative Value):** production scoring follows the **QV funnel** (`docs/mvp/features/quantitative-value.md`); requires multi-period fundamentals for FS-Score YoY deltas.
 
 Terminology reminders:
 
-- “Cheap” means high **EY**, not low P/E—use **EY rank** in issues and code names.
-- “Quality” means high **ROC**, not ESG or subjective moat—use **ROC rank**.
+- **Demo / MF benchmark:** “cheap” = high **EY**; “quality” = high **ROC** — use **EY rank** and **ROC rank** in MF code paths.
+- **Production (Phase 2+):** “cheap” = **EBIT/TEV value pool** membership; “quality” = **FS-Score** — do not use ROC/EY ranks for production portfolio selection.
+- **Combined rank** is **benchmark-only** after Phase 2; production uses **QV funnel rank**.
 - “Value trap” in specs means negative EBIT routed to **review queue**, not a separate score.
 - MVP specs in `docs/mvp/` remain canonical until an ADR or architecture decision supersedes them; update `CONTEXT.md` when `/grill-with-docs` resolves a term conflict.
diff --git a/Dockerfile b/Dockerfile
new file mode 100644
index 0000000..cbceddd
--- /dev/null
+++ b/Dockerfile
@@ -0,0 +1,23 @@
+# Pipeline batch image for ingest, normalize, and score-universe on AWS or locally.
+# Build: make docker-build
+# Run:   docker run --rm -e LAKE_ROOT_URI=file:///lake -v "$PWD/data:/lake" smartwealthai-pipeline score-universe --help
+
+FROM python:3.11-slim-bookworm
+
+WORKDIR /app
+
+ENV POETRY_VERSION=2.1.1 \
+    POETRY_VIRTUALENVS_CREATE=false \
+    POETRY_NO_INTERACTION=1 \
+    LAKE_ROOT_URI=file:///lake
+
+RUN pip install --no-cache-dir "poetry==${POETRY_VERSION}"
+
+COPY pyproject.toml poetry.lock ./
+COPY src ./src
+COPY config ./config
+
+RUN poetry install --only main
+
+ENTRYPOINT ["score-universe"]
+CMD ["--help"]
diff --git a/Makefile b/Makefile
index 665216f..b625ba9 100644
--- a/Makefile
+++ b/Makefile
@@ -1,4 +1,4 @@
-.PHONY: install lint test download-fundamentals
+.PHONY: install lint test download-fundamentals docker-build
 
 install:
 	poetry install --with dev
@@ -10,6 +10,9 @@ lint:
 test:
 	poetry run pytest --cov=smartwealthai --cov-report=term-missing
 
+docker-build:
+	docker build -f Dockerfile -t smartwealthai-pipeline .
+
 # Requires SEC_IDENTITY in the environment. See docs/mvp/guides/download-fundamentals.md
 download-fundamentals:
 	poetry run download-fundamentals --universe dow30
diff --git a/README.md b/README.md
index d69378d..e6ca6f0 100644
--- a/README.md
+++ b/README.md
@@ -62,6 +62,19 @@ Canonical specs: [`docs/mvp/`](docs/mvp/) · Ubiquitous language: [`CONTEXT.md`]
 
     Guide: [`docs/mvp/guides/download-fundamentals.md`](docs/mvp/guides/download-fundamentals.md).
 
+## Cloud foundation (Phase 2a)
+
+Pipeline batch work targets AWS with a **configurable lake root**; the Streamlit dashboard stays local for now.
+
+| Concern | Local dev | AWS integration |
+| --- | --- | --- |
+| Lake root | `LAKE_ROOT_URI=file://…` or `--data-dir data` | `LAKE_ROOT_URI=s3://dev-bucket/prefix/` (I/O slice: file backend in pytest) |
+| PR CI | `make lint` + `make test` on fixtures | No credentials |
+| Ingest-smoke | N/A | [`.github/workflows/ingest-smoke.yml`](.github/workflows/ingest-smoke.yml) via OIDC |
+| Pipeline image | `make docker-build` | ECR/ECS in [#95](https://github.com/JLaborda/SmartWealthAI/issues/95) |
+
+Guide: [`docs/mvp/guides/cloud-foundation.md`](docs/mvp/guides/cloud-foundation.md).
+
 ## Roadmap
 
 See [`docs/mvp/demo-slice.md`](docs/mvp/demo-slice.md) for the **June 30, 2026** delivery target and [`docs/mvp/architecture/architecture.md`](docs/mvp/architecture/architecture.md) for the full MVP north star.
diff --git a/docs/mvp/demo-slice.md b/docs/mvp/demo-slice.md
index 0572334..5430467 100644
--- a/docs/mvp/demo-slice.md
+++ b/docs/mvp/demo-slice.md
@@ -90,9 +90,15 @@ flowchart LR
 
 ## After the demo (phase 2 order)
 
-1. Historical S&P 500 universe + permanent loss filter  
-2. Minimal backtest (annual rebalance, 20 years) — custom pandas/DuckDB loop, not Zipline  
-3. Walk-forward, Monte Carlo, benchmark gate  
-4. Sell-watch + paper trading  
-5. SEC EDGAR normalizer (optional PIT upgrade)  
-6. Quantitative Value metrics (multi-period fundamentals from raw SimFin archives)
+Per [`prds/phase2/prd.md`](prds/phase2/prd.md) — **Quantitative Value is production scoring**; Magic Formula (ROC + EY + combined rank) remains **benchmark-only**.
+
+1. **QV feature spec** — [`quantitative-value.md`](features/quantitative-value.md) (canonical funnel; blocks scoring implementation)  
+2. Multi-period fundamentals + daily prices (FS-Score YoY deltas, backtest NAV)  
+3. Forensic evaluator + Beneish bottom-5% gate (extends [`permanent-loss-filter.md`](features/permanent-loss-filter.md))  
+4. **QV funnel** — EBIT/TEV value decile → FS-Score → ~50-name EW model portfolio  
+5. Light backtest (5–10 years, annual rebalance) vs S&P 500 CW + MF replica  
+6. Cloud pipeline (S3 lake, ECS, MLflow S3 artifacts)  
+7. Sell-watch with QV-adapted triggers + dashboard  
+8. Historical S&P 500 universe (delisted names) + full backtest (20+ years, walk-forward, Sharpe gate)  
+9. SEC EDGAR normalizer (optional PIT upgrade)  
+10. Paper trading + broker (after passing full backtest)
diff --git a/docs/mvp/features/quantitative-value.md b/docs/mvp/features/quantitative-value.md
new file mode 100644
index 0000000..2ee0456
--- /dev/null
+++ b/docs/mvp/features/quantitative-value.md
@@ -0,0 +1,248 @@
+# Feature: Quantitative Value (QV funnel)
+
+## Implementation status
+
+**done** (spec only) — canonical QV feature spec authored in [#86](https://github.com/JLaborda/SmartWealthAI/issues/86). Scoring code deferred to [#89](https://github.com/JLaborda/SmartWealthAI/issues/89)–[#93](https://github.com/JLaborda/SmartWealthAI/issues/93).
+
+## Objective
+
+Define the **production scoring path** for Phase 2+: the *Quantitative Value* funnel from Wesley R. Gray and Tobias Carlisle — forensic hard exclusion, EBIT/TEV value screen, FS-Score quality screen, and a concentrated equal-weight **model portfolio**. The June 30 demo **Magic Formula** path (ROC + EY + combined rank → top 30) remains a **benchmark-only** replica for backtest comparison, not production scoring.
+
+Parent PRD: [`docs/mvp/prds/phase2/prd.md`](../prds/phase2/prd.md).
+
+## MVP scope
+
+- **Sequential funnel** with monotonic shrinkage: each stage only removes names; stage input/output counts are auditable and logged to MLflow.
+- **Stage 1 — Universe pass:** investable tickers from [`universe-construction.md`](universe-construction.md) (sector hard exclusions upstream).
+- **Stage 2 — Forensic hard exclusion:** evaluate all rules in [`permanent-loss-filter.md`](permanent-loss-filter.md) plus **Beneish M-Score** with a QVAL-style **bottom-5% cross-sectional gate** per forensic model among survivors. Emit `exclude` or `pass` with `rule_id`, `rule_version`, `triggered_value`, `threshold`, `explanation`.
+- **Stage 3 — Value screen:** rank forensic survivors by **EBIT/TEV** descending (same EV definition as [`cheap-stocks.md`](cheap-stocks.md)); keep the top **decile (~10%)**, configurable via `value_decile_fraction`.
+- **Stage 4 — Quality screen:** compute **FS-Score** (Gray/Carlisle 10-component variant) on the value pool; rank by FS-Score descending; keep the top **N** names up to the portfolio cap (default **50**).
+- **Stage 5 — Model portfolio:** equal-weight long-only holdings; **market-cap tie-break** on rank ties (ascending, same convention as demo MF).
+- **Configuration:** versioned YAML under `config/quantitative_value/` (funnel parameters) and `config/permanent_loss/` (forensic thresholds, Beneish coefficients).
+- **Outputs:** per-stage parquet tables, final `model_portfolio` rows, MLflow metrics (`qv_stage_*_count`, config hashes, `git_sha`).
+- **Explainability:** dashboard shows stage membership, EBIT/TEV rank, FS-Score components, and forensic pass/exclusion reasons per ticker.
+
+## Out of MVP scope
+
+- Implementing forensic evaluator, FS-Score calculator, QV orchestrator, or scoring CLI wiring ([#89](https://github.com/JLaborda/SmartWealthAI/issues/89)–[#91](https://github.com/JLaborda/SmartWealthAI/issues/91)).
+- Multi-period ETL and daily prices ([#87](https://github.com/JLaborda/SmartWealthAI/issues/87), [#88](https://github.com/JLaborda/SmartWealthAI/issues/88)) — prerequisite data work, not part of this spec's implementation slice.
+- Cloud deploy, ECS, S3 lake root ([#94](https://github.com/JLaborda/SmartWealthAI/issues/94), [#95](https://github.com/JLaborda/SmartWealthAI/issues/95)).
+- Changing demo Magic Formula scoring behavior (ROC/EY/combined rank path preserved for benchmark).
+- Moat "pre-flight checklist" and full forensic model zoo beyond Beneish + spec distress rules.
+- Score-weighted and risk-parity portfolio weighting (equal-weight only in Phase 2a).
+- Paper trading, broker execution, auto-execution of sell signals.
+
+## Inputs
+
+| Input | Source | Notes |
+| --- | --- | --- |
+| Universe pass list | `curated/universe` | `pass` rows only; banks/insurers/utilities already excluded. |
+| PIT fundamentals (multi-period) | `curated/fundamentals` | `as_of_date <= run_date`; annual/quarterly history required for FS-Score YoY deltas ([#87](https://github.com/JLaborda/SmartWealthAI/issues/87)). |
+| Run-date prices | `curated/prices` | Market cap and EV components at `run_date`. |
+| Forensic rules + Beneish config | `config/permanent_loss/` | Thresholds and coefficient sets are versioned. |
+| QV funnel config | `config/quantitative_value/` | `value_decile_fraction`, `portfolio_size`, `formula_version`, tie-break policy. |
+| Run date | Pipeline parameter | Decision date for PIT queries and cross-sectional ranks. |
+| Permanent loss exclusions (reference) | [`permanent-loss-filter.md`](permanent-loss-filter.md) | Bankruptcy/distress + fraud rules; extended here with Beneish. |
+
+## Outputs
+
+| Output | Path / target |
+| --- | --- |
+| Stage counts (MLflow) | `qv_stage_universe_count`, `qv_stage_forensic_pass_count`, `qv_stage_value_pool_count`, `qv_stage_quality_pool_count`, `qv_stage_portfolio_count` |
+| Forensic exclusions | `curated/permanent_loss/run_date=<YYYY-MM-DD>/exclusions.parquet` — columns `cik, ticker, subfilter, rule_id, rule_version, triggered_value, threshold, percentile, as_of_date, explanation` |
+| Value pool scores | `curated/scores/qv_value/run_date=<YYYY-MM-DD>/scores.parquet` — `cik, ticker, ebit, ev, ebit_tev, value_rank, formula_version, as_of_date` |
+| Quality scores | `curated/scores/qv_quality/run_date=<YYYY-MM-DD>/scores.parquet` — `cik, ticker, fs_score, fs_*` component columns, `formula_version`, `as_of_date` |
+| Funnel audit | `curated/scores/qv_funnel/run_date=<YYYY-MM-DD>/funnel.parquet` — per-ticker stage reached, exclusion reason if any |
+| Model portfolio (QV) | `curated/portfolio/qv/run_date=<YYYY-MM-DD>/portfolio.parquet` — `cik, ticker, qv_funnel_rank, fs_score, ebit_tev, weight, formula_version` |
+| Review queue | `curated/issues/run_date=<YYYY-MM-DD>/qv.parquet` — missing FS-Score or forensic inputs |
+
+## Funnel stages
+
+| Stage | Name | Action | Typical shrinkage |
+| --- | --- | --- | --- |
+| 0 | Universe | Sector-filtered investable set | Baseline count |
+| 1 | Forensic | Hard `exclude` on any distress/fraud rule **or** Beneish bottom 5% | Largest drop |
+| 2 | Value | Top decile by EBIT/TEV among survivors | ~90% removed |
+| 3 | Quality | FS-Score rank within value pool; select top N by score | To ~50 names |
+| 4 | Portfolio | Equal-weight; market-cap tie-break on rank ties | Final holdings |
+
+**Monotonicity contract:** a name excluded at stage *k* never appears in stages *k+1* … portfolio. Stage counts must satisfy `count[k] >= count[k+1]` for all adjacent stages.
+
+## Forensic stage (extends permanent loss filter)
+
+Forensic evaluation runs **before** any value or quality score. It **references and extends** [`permanent-loss-filter.md`](permanent-loss-filter.md):
+
+| Source | Rules | MVP behavior |
+| --- | --- | --- |
+| Bankruptcy / distress | `BK_ALTMAN_Z`, `BK_INT_COVERAGE`, `BK_NETDEBT_EBITDA`, `BK_NEGATIVE_EQUITY`, `BK_DELISTED` | Hard `exclude` if any rule fires |
+| Fraud | `FRD_RESTATEMENT_RECENT`, `FRD_AUDITOR_CHANGE_REPEATED`, `FRD_LATE_FILER` | Hard `exclude` if any rule fires; EDGAR-dependent rules may be `unavailable` until wired |
+| Manipulation (QV) | `FRD_BENEISH_M` | Compute Beneish M-Score; **exclude names in the bottom 5%** of the cross-sectional M-Score distribution among forensic-stage candidates |
+
+### Beneish M-Score bottom-5% gate
+
+- **Rule id:** `FRD_BENEISH_M`
+- **Config:** `config/permanent_loss/beneish.yaml` (`rule_version`, coefficient set, minimum input coverage)
+- **Gate:** among companies with a computable M-Score at `run_date`, exclude those at or below the **5th percentile** (QVAL-style safety screen per forensic model).
+- **Audit columns:** `rule_id`, `rule_version`, `triggered_value` (M-Score), `threshold` (5th percentile cutoff), `percentile`, `explanation`
+- **Missing inputs:** route to review queue; do not silently pass (same policy as permanent-loss spec).
+
+Distress and fraud rules from the permanent-loss spec retain their existing thresholds in `config/permanent_loss/`; Beneish coefficients and percentile gate live in the same namespace with separate version ids.
+
+## Value screen (EBIT/TEV)
+
+Production **cheapness** is membership in the **EBIT/TEV value pool**, not MF **EY rank** alone.
+
+```
+EBIT/TEV = EBIT / Enterprise Value
+EV = Market Cap + Total Debt + Preferred Equity + Minority Interest - Cash
+```
+
+- Same `EBIT` and EV component definitions as [`cheap-stocks.md`](cheap-stocks.md) (`formula_version` shared where components overlap).
+- Cross-sectional rank **within forensic survivors** by descending EBIT/TEV.
+- Keep top `value_decile_fraction` (default **0.10** ≈ decile).
+- `EBIT <= 0` or invalid EV → review queue, not value pool.
+- Output metric is **EBIT/TEV** and `value_rank`; this path is separate from MF `ey_rank`.
+
+## FS-Score (Gray/Carlisle variant)
+
+Production **quality** is the **FS-Score** composite on the value pool, not ROC rank alone. Ten binary components (1 = good, 0 = bad); sum to integer **0–10**. Reference: Gray & Carlisle, *Quantitative Value* (2013), Ch. 6; [Alpha Architect FS-Score article](https://alphaarchitect.com/2015/05/value-investing-research-simple-methods-to-improve-the-piotroski-f-score/).
+
+`formula_version` (initial: `fs_v1`) is recorded on every scored row.
+
+### Current profitability
+
+| Component id | Definition | Score = 1 when |
+| --- | --- | --- |
+| `FS_ROA` | ROA = net income before extraordinary items / total assets (most recent fiscal year) | ROA > 0 |
+| `FS_FCFTA` | FCFTA = free cash flow / total assets | FCFTA > 0 |
+| `FS_ACCRUAL` | Accrual quality signal | FCFTA > ROA |
+
+### Stability
+
+| Component id | Definition | Score = 1 when |
+| --- | --- | --- |
+| `FS_ΔLEVER` | Change in long-term debt / total assets | Leverage ratio **decreased** YoY |
+| `FS_ΔLIQUID` | Change in current ratio (current assets / current liabilities) | Liquidity ratio **increased** YoY |
+| `FS_NEQISS` | Net equity issuance = repurchases − issuances | Repurchases **exceed** issuances |
+
+### Recent operational improvements
+
+| Component id | Definition | Score = 1 when |
+| --- | --- | --- |
+| `FS_ΔROA` | Current ROA − prior ROA | ΔROA > 0 |
+| `FS_ΔFCFTA` | Current FCFTA − prior FCFTA | ΔFCFTA > 0 |
+| `FS_ΔMARGIN` | Current gross margin − prior gross margin | ΔMARGIN > 0 |
+| `FS_ΔTURN` | Current asset turnover − prior asset turnover | ΔTURN > 0 |
+
+**FS-Score** = sum of all ten components. **QV funnel rank** within the value pool orders by FS-Score descending; ties break by ascending market cap (configurable in `config/quantitative_value/tie_break.yaml`).
+
+Missing prior-year inputs for a component → component scores 0 and row is flagged in review queue if coverage falls below configured minimum.
+
+## Model portfolio (QV)
+
+| Topic | Decision (closed) |
+| --- | --- |
+| Default size | **50** names |
+| Configurable cap | `portfolio_size` in `config/quantitative_value/funnel.yaml` |
+| Weighting | Equal-weight only (Phase 2a) |
+| Selection | Top N by FS-Score within value pool |
+| Tie-break | Ascending market cap on rank ties |
+| Long/short | Long-only |
+
+This supersedes the demo slice top-30 MF portfolio for **production** scoring only; the MF replica may use a different N for benchmark comparison.
+
+## Key interfaces (spec contracts)
+
+These are **documentation contracts** for downstream implementation issues; no code in this issue.
+
+### QV funnel orchestrator
+
+- **Input:** `run_date`, lake root, QV + forensic config hashes.
+- **Behavior:** run stages 0→4 sequentially; delegate to forensic evaluator, value ranker, FS-Score ranker, portfolio constructor; emit stage-count metrics.
+- **Output:** `ScoringResult`-like object with per-stage tables, final portfolio, MLflow-ready metrics.
+
+### Forensic evaluator
+
+- **Input:** universe pass list + PIT fundamentals (+ optional filing flags).
+- **Output:** `exclude` or `pass` per `(cik, run_date)` with full audit columns.
+
+### FS-Score calculator
+
+- **Input:** multi-period PIT income/balance/cashflow for one ticker at `decision_date`.
+- **Output:** component dict, total 0–10, `formula_version`.
+
+### Value metrics
+
+- Reuse EV/EBIT building blocks from cheap-stocks module; separate production path from MF `EY rank`.
+
+### Magic Formula benchmark path
+
+- Preserve ROC + EY + **combined rank** unchanged; invoke only from backtest benchmark builder and dedicated CI tests.
+
+## Configuration namespace
+
+| Path | Contents |
+| --- | --- |
+| `config/quantitative_value/funnel.yaml` | `portfolio_size` (default 50), `value_decile_fraction` (default 0.10), stage ordering |
+| `config/quantitative_value/fs_score.yaml` | `formula_version`, minimum input coverage, fiscal-year alignment rules |
+| `config/quantitative_value/tie_break.yaml` | Market-cap tie-break policy |
+| `config/permanent_loss/*.yaml` | Distress/fraud thresholds (existing) + `beneish.yaml` (coefficients, bottom-percentile gate) |
+
+Every MLflow run logs config file hashes and `formula_version` values.
+
+## Mermaid diagram
+
+```mermaid
+flowchart TD
+    Uni["Stage 0: Universe pass"] --> Forensic["Stage 1: Forensic evaluator"]
+    Forensic --> Distress{"Distress / fraud rule fired?"}
+    Distress -->|Yes| Excl["Hard exclude + audit row"]
+    Distress -->|No| Beneish{"Beneish M-Score bottom 5%?"}
+    Beneish -->|Yes| Excl
+    Beneish -->|No| Value["Stage 2: EBIT/TEV rank → top decile"]
+    Value --> Quality["Stage 3: FS-Score on value pool"]
+    Quality --> Port["Stage 4: Top N EW portfolio (~50)"]
+    Excl --> Audit["exclusions.parquet + MLflow metrics"]
+    Port --> Out["portfolio.parquet + funnel audit"]
+    Uni -.-> MLflow["qv_stage_*_count"]
+    Forensic -.-> MLflow
+    Value -.-> MLflow
+    Quality -.-> MLflow
+    Port -.-> MLflow
+```
+
+## Expected flow
+
+1. Load universe pass list for `run_date`; log `qv_stage_universe_count`.
+2. Run forensic evaluator (permanent-loss rules + Beneish bottom-5% gate); write exclusions; log `qv_stage_forensic_pass_count`.
+3. Rank forensic survivors by EBIT/TEV; keep top decile; log `qv_stage_value_pool_count`.
+4. Compute FS-Score for value-pool names; rank by total score; select top `portfolio_size`; log `qv_stage_quality_pool_count`.
+5. Build equal-weight portfolio with market-cap tie-break; log `qv_stage_portfolio_count`.
+6. Persist parquet artifacts and MLflow run (params, stage metrics, portfolio artifact, `git_sha`).
+
+## Acceptance criteria
+
+- [x] Spec exists with objective, MVP scope, inputs/outputs, funnel stages, Mermaid diagram, acceptance criteria, and risks ([#86](https://github.com/JLaborda/SmartWealthAI/issues/86)).
+- [x] FS-Score documents all **10 binary components** with Gray/Carlisle definitions and `formula_version`.
+- [x] Forensic stage references [`permanent-loss-filter.md`](permanent-loss-filter.md) and documents Beneish **bottom-5%** gate with audit columns.
+- [x] Portfolio size default **~50** recorded as a closed decision with configurable cap.
+- [ ] Funnel monotonic shrinkage: each stage only removes names; stage counts auditable in MLflow (implementation [#91](https://github.com/JLaborda/SmartWealthAI/issues/91)).
+- [ ] Forensic hard exclusion runs before any value/quality score (implementation [#89](https://github.com/JLaborda/SmartWealthAI/issues/89)).
+- [ ] Value screen keeps top decile by EBIT/TEV among survivors (implementation [#90](https://github.com/JLaborda/SmartWealthAI/issues/90)).
+- [ ] Quality screen ranks FS-Score within value pool; portfolio cap applied last (implementation [#91](https://github.com/JLaborda/SmartWealthAI/issues/91)).
+- [ ] MF **combined rank** remains benchmark-only; demo ROC/EY path unchanged on fixtures (regression in CI).
+- [ ] Same `(universe, run_date, config hashes)` produces byte-identical funnel outputs when implementation lands.
+
+## Open questions
+
+- Beneish M-Score: use classic 8-variable model or include optional 9th/10th variables when data exists? **Recommendation:** start with 8-variable `M-Score` per Beneish (1999); document coefficient set in `beneish.yaml`; extend in a new `rule_version` if coverage improves.
+- FS-Score fiscal alignment: match on fiscal year-end or trailing four quarters? **Recommendation:** fiscal year pairs for YoY deltas; flag mismatched fiscal calendars in review queue.
+- Value decile: fixed 10% fraction vs fixed count? **Recommendation:** fraction (`value_decile_fraction = 0.10`) so pool scales with universe size.
+
+## Risks
+
+- **Missing multi-period data** shrinks FS-Score coverage; depends on [#87](https://github.com/JLaborda/SmartWealthAI/issues/87) landing first.
+- **Beneish false positives** on growth firms with high DSRI; bottom-5% gate mitigates but does not eliminate FP review burden.
+- **Survivorship bias** in Phase 2a light backtest (current SimFin US universe) — must be labeled in reports per PRD.
+- **Compute cost:** full funnel × rebalance dates in backtests; plan DuckDB pushdown or materialized stage tables early.
+- **Vocabulary drift:** production terms (FS-Score, EBIT/TEV value pool, QV funnel rank) must stay distinct from MF benchmark terms in code and `CONTEXT.md`.
diff --git a/docs/mvp/guides/cloud-foundation.md b/docs/mvp/guides/cloud-foundation.md
new file mode 100644
index 0000000..cf8f6bb
--- /dev/null
+++ b/docs/mvp/guides/cloud-foundation.md
@@ -0,0 +1,97 @@
+# Cloud foundation (Phase 2a)
+
+Operator guide for **lake root URI**, the pipeline Docker image, and the **ingest-smoke** GitHub Actions workflow. Implements GitHub issue [#94](https://github.com/JLaborda/SmartWealthAI/issues/94). ECS Fargate deploy is [#95](https://github.com/JLaborda/SmartWealthAI/issues/95).
+
+## Lake root URI
+
+All ingest and scoring CLIs resolve storage from a single **lake root**:
+
+| Source | Precedence |
+| --- | --- |
+| `--lake-root-uri` | 1 (highest) |
+| `--data-dir` | 2 |
+| `LAKE_ROOT_URI` env var | 3 |
+| `SMARTWEALTHAI_DATA_DIR` env var | 4 |
+| default `data/` | 5 |
+
+Supported URI forms:
+
+- `file:///absolute/path` or bare local path (`data`, `/tmp/lake`)
+- `s3://bucket/prefix/` (parsed for cloud workflows; parquet I/O in this slice is **file-backend only**)
+
+Example (local mirror):
+
+```bash
+export LAKE_ROOT_URI="file://$(pwd)/data"
+poetry run score-universe --run-date 2026-06-18
+```
+
+Backward compatible:
+
+```bash
+poetry run score-universe --data-dir data --run-date 2026-06-18
+```
+
+## Pipeline Docker image
+
+Build and smoke-test locally (no AWS):
+
+```bash
+make docker-build
+docker run --rm smartwealthai-pipeline --help
+docker run --rm \
+  -e LAKE_ROOT_URI=file:///lake \
+  -v "$(pwd)/tests/fixtures/lake:/lake:ro" \
+  smartwealthai-pipeline \
+  --run-date 2026-06-18 --portfolio-size 3 --quiet
+```
+
+The image entrypoint is `score-universe`. Override the command for other CLIs by installing Poetry in a dev image or extending the Dockerfile entrypoint in [#95](https://github.com/JLaborda/SmartWealthAI/issues/95).
+
+## Ingest-smoke workflow
+
+Workflow: [`.github/workflows/ingest-smoke.yml`](../../.github/workflows/ingest-smoke.yml)
+
+- Triggers: manual `workflow_dispatch` and weekly cron (Mondays 06:00 UTC)
+- Uses GitHub Environment **`dev`** and **OIDC** (no long-lived AWS access keys)
+- Writes `raw/_smoke/ingest-smoke.txt` to the dev bucket and validates with `head-object`
+
+### Human prerequisites (one-time AWS setup)
+
+If buckets and OIDC are not yet provisioned in your AWS account:
+
+1. **S3 dev bucket** — e.g. `smartwealthai-dev-lake` with optional prefix `data/`. Same zone layout as local: `raw/`, `curated/`, `curated/issues/`.
+2. **GitHub OIDC provider** in IAM (issuer `token.actions.githubusercontent.com`, audience `sts.amazonaws.com`).
+3. **IAM role** trusted by the repo and `environment:dev`, with `s3:PutObject`, `s3:GetObject`, `s3:ListBucket` on the dev bucket prefix.
+4. **GitHub Environment `dev`** repository variables:
+
+   | Variable | Example |
+   | --- | --- |
+   | `AWS_ROLE_ARN` | `arn:aws:iam::123456789012:role/github-actions-dev` |
+   | `AWS_REGION` | `us-east-1` |
+   | `DEV_LAKE_BUCKET` | `smartwealthai-dev-lake` |
+   | `DEV_LAKE_PREFIX` | `data/` (or empty string) |
+
+5. **Runtime secrets (ECS, #95)** — `SIMFIN_API_KEY` in AWS Secrets Manager; not required for ingest-smoke marker upload.
+
+Terraform or CloudFormation for the above is out of scope for #94; document and provision manually or in a future infra issue.
+
+### Run ingest-smoke manually
+
+GitHub → Actions → **Ingest smoke (dev S3)** → Run workflow (environment: `dev`).
+
+## Three-tier testing model
+
+| Tier | What runs | AWS / network |
+| --- | --- | --- |
+| **PR CI** | `make lint`, `pytest` on fixtures | None |
+| **Ingest-smoke** | OIDC S3 marker + lake URI parse | Dev bucket only |
+| **Deploy (#95)** | ECR push, ECS Fargate pipeline | Dev/prod on merge |
+
+PR CI must stay hermetic: no SimFin, yfinance, or live S3.
+
+## Phase 2a scope
+
+- **Pipeline on AWS** (batch ingest + score) — this guide + Docker image + ingest-smoke
+- **Dashboard local** — `poetry run run-dashboard` reads `SMARTWEALTHAI_DATA_DIR` or local `data/`
+- Full scheduled ECS deploy, MLflow on S3, and Secrets Manager task injection: [#95](https://github.com/JLaborda/SmartWealthAI/issues/95)
diff --git a/src/smartwealthai/build_universe.py b/src/smartwealthai/build_universe.py
index d22d110..15997ac 100644
--- a/src/smartwealthai/build_universe.py
+++ b/src/smartwealthai/build_universe.py
@@ -8,18 +8,13 @@
 import click
 from click.testing import CliRunner
 
+from smartwealthai.cli_lake import lake_root_options, resolve_cli_data_dir
 from smartwealthai.simfin_industry_exclusions import REFERENCE_PATH
 from smartwealthai.universe_builder import build_universe
 
 
 @click.command(context_settings={"help_option_names": ["-h", "--help"]})
-@click.option(
-    "--data-dir",
-    type=click.Path(path_type=Path, file_okay=False),
-    default=Path("data"),
-    show_default=True,
-    help="Data lake root.",
-)
+@lake_root_options
 @click.option(
     "--run-date",
     type=click.DateTime(formats=["%Y-%m-%d"]),
@@ -40,16 +35,22 @@
     help="Industry exclusions reference CSV.",
 )
 def main(
-    data_dir: Path,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
     run_date: datetime,
     snapshot_date: datetime | None,
     exclusions_file: Path,
 ) -> None:
     """Build curated universe and exclusion artifacts for one run date."""
+    try:
+        lake_path = resolve_cli_data_dir(lake_root_uri=lake_root_uri, data_dir=data_dir)
+    except ValueError as exc:
+        raise click.ClickException(str(exc)) from exc
+
     decision_date = run_date.date()
     snapshot = snapshot_date.date() if snapshot_date is not None else decision_date
     result = build_universe(
-        data_dir,
+        lake_path,
         run_date=decision_date,
         snapshot_date=snapshot,
         exclusions_path=exclusions_file,
diff --git a/src/smartwealthai/cli_lake.py b/src/smartwealthai/cli_lake.py
new file mode 100644
index 0000000..2f0aa81
--- /dev/null
+++ b/src/smartwealthai/cli_lake.py
@@ -0,0 +1,57 @@
+"""Shared lake root resolution for pipeline CLIs."""
+
+from __future__ import annotations
+
+import os
+from collections.abc import Callable
+from pathlib import Path
+from typing import Any
+
+import click
+
+from smartwealthai.lake_root import LakeRoot, resolve_lake_root
+
+
+def lake_root_options(fn: Callable[..., Any]) -> Callable[..., Any]:
+    """Attach ``--lake-root-uri`` and ``--data-dir`` options to a Click command."""
+    fn = click.option(
+        "--lake-root-uri",
+        envvar="LAKE_ROOT_URI",
+        default=None,
+        help="Lake root URI (file:// or s3://). Overrides --data-dir.",
+    )(fn)
+    fn = click.option(
+        "--data-dir",
+        type=click.Path(path_type=Path, file_okay=False),
+        default=None,
+        help="Local data lake root (backward compatible; default data/).",
+    )(fn)
+    return fn
+
+
+def resolve_cli_lake_root(
+    *,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
+) -> LakeRoot:
+    """Resolve lake root with CLI/env precedence.
+
+    Order: ``--lake-root-uri`` (includes ``LAKE_ROOT_URI`` via Click) →
+    ``--data-dir`` → ``SMARTWEALTHAI_DATA_DIR`` → ``data``.
+    """
+    if lake_root_uri:
+        return resolve_lake_root(lake_root_uri)
+    if data_dir is not None:
+        return resolve_lake_root(str(data_dir))
+    if uri := os.environ.get("SMARTWEALTHAI_DATA_DIR"):
+        return resolve_lake_root(uri)
+    return resolve_lake_root("data")
+
+
+def resolve_cli_data_dir(
+    *,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
+) -> Path:
+    """Return a local :class:`Path` for file-backend pipeline stages."""
+    return resolve_cli_lake_root(lake_root_uri=lake_root_uri, data_dir=data_dir).as_path()
diff --git a/src/smartwealthai/compute_metrics.py b/src/smartwealthai/compute_metrics.py
index 16a9a58..ffd444f 100644
--- a/src/smartwealthai/compute_metrics.py
+++ b/src/smartwealthai/compute_metrics.py
@@ -8,6 +8,7 @@
 import click
 from click.testing import CliRunner
 
+from smartwealthai.cli_lake import lake_root_options, resolve_cli_data_dir
 from smartwealthai.magic_formula_metrics import MetricsResult
 from smartwealthai.pit_fundamentals import MetricsInputError, compute_metrics_for_ticker
 
@@ -18,13 +19,7 @@
     required=True,
     help="Ticker symbol (e.g. AAPL).",
 )
-@click.option(
-    "--data-dir",
-    type=click.Path(path_type=Path, file_okay=False),
-    default=Path("data"),
-    show_default=True,
-    help="Data lake root.",
-)
+@lake_root_options
 @click.option(
     "--as-of-date",
     type=click.DateTime(formats=["%Y-%m-%d"]),
@@ -32,12 +27,22 @@
     show_default="today (UTC)",
     help="Decision date (PIT fundamentals and price snapshot).",
 )
-def main(ticker: str, data_dir: Path, as_of_date: datetime) -> None:
+def main(
+    ticker: str,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
+    as_of_date: datetime,
+) -> None:
     """Compute ROC and EY for one ticker using curated fundamentals and prices."""
+    try:
+        lake_path = resolve_cli_data_dir(lake_root_uri=lake_root_uri, data_dir=data_dir)
+    except ValueError as exc:
+        raise click.ClickException(str(exc)) from exc
+
     decision_date = as_of_date.date()
     try:
         result = compute_metrics_for_ticker(
-            data_dir,
+            lake_path,
             ticker=ticker.upper(),
             as_of_date=decision_date,
         )
diff --git a/src/smartwealthai/download_fundamentals.py b/src/smartwealthai/download_fundamentals.py
index b027cd9..c6f25c2 100644
--- a/src/smartwealthai/download_fundamentals.py
+++ b/src/smartwealthai/download_fundamentals.py
@@ -24,6 +24,7 @@
 
 import click
 
+from smartwealthai.cli_lake import lake_root_options, resolve_cli_data_dir
 from smartwealthai.edgartools_client import EdgartoolsClientError, download_statements
 from smartwealthai.lake_paths import (
     STATEMENT_NAMES,
@@ -189,13 +190,7 @@ def run_download(
     type=click.Path(path_type=Path, dir_okay=False),
     help="Path to a universe CSV with columns ticker,cik.",
 )
-@click.option(
-    "--data-dir",
-    type=click.Path(path_type=Path, file_okay=False),
-    default=Path("data"),
-    show_default=True,
-    help="Data lake root.",
-)
+@lake_root_options
 @click.option(
     "--periods",
     type=int,
@@ -217,18 +212,24 @@ def run_download(
 def main(
     universe: str | None,
     universe_file: Path | None,
-    data_dir: Path,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
     periods: int,
     as_of_date: datetime | None,
     force: bool,
 ) -> None:
     """Download SEC companyfacts and edgartools annual statements for a universe."""
+    try:
+        lake_path = resolve_cli_data_dir(lake_root_uri=lake_root_uri, data_dir=data_dir)
+    except ValueError as exc:
+        raise click.ClickException(str(exc)) from exc
+
     snapshot_date = as_of_date.date() if as_of_date is not None else None
     raise SystemExit(
         run_download(
             universe=universe,
             universe_file=universe_file,
-            data_dir=data_dir,
+            data_dir=lake_path,
             periods=periods,
             as_of_date=snapshot_date,
             force=force,
diff --git a/src/smartwealthai/download_prices.py b/src/smartwealthai/download_prices.py
index c023e82..5eb2c14 100644
--- a/src/smartwealthai/download_prices.py
+++ b/src/smartwealthai/download_prices.py
@@ -10,6 +10,7 @@
 import click
 from click.testing import CliRunner
 
+from smartwealthai.cli_lake import lake_root_options, resolve_cli_data_dir
 from smartwealthai.lake_paths import price_ingest_errors_path
 from smartwealthai.price_ingest import run_price_ingest
 
@@ -17,13 +18,7 @@
 
 
 @click.command(context_settings={"help_option_names": ["-h", "--help"]})
-@click.option(
-    "--data-dir",
-    type=click.Path(path_type=Path, file_okay=False),
-    default=Path("data"),
-    show_default=True,
-    help="Data lake root.",
-)
+@lake_root_options
 @click.option(
     "--run-date",
     type=click.DateTime(formats=["%Y-%m-%d"]),
@@ -42,19 +37,25 @@
     help="Rebuild curated snapshot even when it already exists.",
 )
 def main(
-    data_dir: Path,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
     run_date: datetime,
     snapshot_date: datetime | None,
     force: bool,
 ) -> None:
     """Build curated run-date prices from SimFin bulk shareprices/latest."""
+    try:
+        lake_path = resolve_cli_data_dir(lake_root_uri=lake_root_uri, data_dir=data_dir)
+    except ValueError as exc:
+        raise click.ClickException(str(exc)) from exc
+
     logging.basicConfig(level=logging.INFO, format="%(levelname)s %(message)s")
     decision_date = run_date.date()
     snapshot = snapshot_date.date() if snapshot_date is not None else decision_date
 
     try:
         ingest_run = run_price_ingest(
-            data_dir=data_dir,
+            data_dir=lake_path,
             run_date=decision_date,
             snapshot_date=snapshot,
             force=force,
@@ -77,7 +78,7 @@ def main(
             {"ticker": ticker, "error": "no SimFin share price on or before run_date"}
             for ticker in ingest_run.missing_tickers
         ]
-        error_file = price_ingest_errors_path(data_dir, run_date=decision_date)
+        error_file = price_ingest_errors_path(lake_path, run_date=decision_date)
         error_file.parent.mkdir(parents=True, exist_ok=True)
         error_file.write_text(json.dumps(payload, indent=2))
         click.echo(f"Wrote error summary: {error_file}")
diff --git a/src/smartwealthai/download_simfin.py b/src/smartwealthai/download_simfin.py
index 96d065d..8738601 100644
--- a/src/smartwealthai/download_simfin.py
+++ b/src/smartwealthai/download_simfin.py
@@ -28,6 +28,7 @@
 import click
 from click.testing import CliRunner
 
+from smartwealthai.cli_lake import lake_root_options, resolve_cli_data_dir
 from smartwealthai.lake_paths import simfin_bulk_path, simfin_errors_path
 from smartwealthai.simfin_client import configure_simfin, fetch_dataset_csv
 
@@ -181,13 +182,7 @@ def run_download(
 
 
 @click.command(context_settings={"help_option_names": ["-h", "--help"]})
-@click.option(
-    "--data-dir",
-    type=click.Path(path_type=Path, file_okay=False),
-    default=Path("data"),
-    show_default=True,
-    help="Data lake root.",
-)
+@lake_root_options
 @click.option(
     "--as-of-date",
     type=click.DateTime(formats=["%Y-%m-%d"]),
@@ -207,16 +202,22 @@ def run_download(
     help="Re-download and overwrite lake copies regardless of age.",
 )
 def main(
-    data_dir: Path,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
     as_of_date: datetime | None,
     refresh_days: int,
     force: bool,
 ) -> None:
     """Download SimFin bulk US fundamentals into the raw lake."""
+    try:
+        lake_path = resolve_cli_data_dir(lake_root_uri=lake_root_uri, data_dir=data_dir)
+    except ValueError as exc:
+        raise click.ClickException(str(exc)) from exc
+
     snapshot_date = as_of_date.date() if as_of_date is not None else datetime.now(UTC).date()
     raise SystemExit(
         run_download(
-            data_dir=data_dir,
+            data_dir=lake_path,
             as_of_date=snapshot_date,
             refresh_days=refresh_days,
             force=force,
diff --git a/src/smartwealthai/lake_root.py b/src/smartwealthai/lake_root.py
new file mode 100644
index 0000000..1a72c6d
--- /dev/null
+++ b/src/smartwealthai/lake_root.py
@@ -0,0 +1,107 @@
+"""Lake root URI resolution and storage-backend I/O."""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import date
+from pathlib import Path
+from typing import Literal
+from urllib.parse import urlparse
+
+import pandas as pd
+
+from smartwealthai import lake_paths
+
+Backend = Literal["file", "s3"]
+
+
+@dataclass(frozen=True)
+class LakeRoot:
+    """Configurable data lake root with zone-relative path builders."""
+
+    backend: Backend
+    uri: str
+    _path: Path | None = None
+    s3_bucket: str | None = None
+    s3_prefix: str | None = None
+
+    def as_path(self) -> Path:
+        """Return the local filesystem root (file backend only)."""
+        if self.backend != "file":
+            msg = f"S3 lake root has no local Path; use lake I/O helpers: {self.uri}"
+            raise TypeError(msg)
+        if self._path is None:
+            msg = f"File lake root is missing a path: {self.uri}"
+            raise ValueError(msg)
+        return self._path
+
+    def curated_portfolio_path(self, *, run_date: date) -> Path:
+        """Build the curated portfolio parquet path for a run date."""
+        if self.backend != "file":
+            msg = "curated_portfolio_path is implemented for the file backend in this slice"
+            raise NotImplementedError(msg)
+        return lake_paths.curated_portfolio_path(self.as_path(), run_date=run_date)
+
+    def write_parquet(self, path: Path, frame: pd.DataFrame) -> None:
+        """Write a parquet artifact under the lake root."""
+        if self.backend != "file":
+            msg = "S3 parquet writes are not implemented in this slice; use file:// locally"
+            raise NotImplementedError(msg)
+        path.parent.mkdir(parents=True, exist_ok=True)
+        frame.to_parquet(path, index=False)
+
+    def read_parquet(self, path: Path) -> pd.DataFrame:
+        """Read a parquet artifact from the lake root."""
+        if self.backend != "file":
+            msg = "S3 parquet reads are not implemented in this slice; use file:// locally"
+            raise NotImplementedError(msg)
+        return pd.read_parquet(path)
+
+
+def resolve_lake_root(uri: str | None) -> LakeRoot:
+    """Parse a lake root URI into a :class:`LakeRoot`.
+
+    Supported forms:
+    - ``file:///absolute/path`` or ``file://relative/path``
+    - bare local path (``data``, ``/tmp/lake``)
+    - ``s3://bucket/prefix`` (parsed for cloud workflows; file I/O stays local here)
+    """
+    if uri is None or not str(uri).strip():
+        msg = "Lake root URI is required"
+        raise ValueError(msg)
+
+    raw = str(uri).strip()
+    if raw.startswith("s3://"):
+        return _resolve_s3(raw)
+    if raw.startswith("file://"):
+        return _resolve_file(raw)
+    if "://" in raw:
+        scheme = urlparse(raw).scheme
+        msg = f"Unsupported lake root URI scheme: {scheme!r}"
+        raise ValueError(msg)
+    return _resolve_file(raw)
+
+
+def _resolve_file(uri: str) -> LakeRoot:
+    if uri.startswith("file://"):
+        parsed = urlparse(uri)
+        if parsed.netloc:
+            path = Path(f"/{parsed.netloc}{parsed.path}")
+        else:
+            path = Path(parsed.path)
+    else:
+        path = Path(uri)
+    resolved = path.expanduser().resolve()
+    return LakeRoot(backend="file", uri=uri, _path=resolved)
+
+
+def _resolve_s3(uri: str) -> LakeRoot:
+    parsed = urlparse(uri)
+    bucket = parsed.netloc
+    if not bucket:
+        msg = f"Invalid S3 lake root URI: {uri!r}"
+        raise ValueError(msg)
+    prefix = parsed.path.lstrip("/")
+    if prefix and not prefix.endswith("/"):
+        prefix = f"{prefix}/"
+    return LakeRoot(backend="s3", uri=uri, s3_bucket=bucket, s3_prefix=prefix or "")
diff --git a/src/smartwealthai/normalize_simfin.py b/src/smartwealthai/normalize_simfin.py
index 4190e3e..72e2de0 100644
--- a/src/smartwealthai/normalize_simfin.py
+++ b/src/smartwealthai/normalize_simfin.py
@@ -9,6 +9,7 @@
 import pandas as pd
 from click.testing import CliRunner
 
+from smartwealthai.cli_lake import lake_root_options, resolve_cli_data_dir
 from smartwealthai.lake_paths import curated_universe_path
 from smartwealthai.simfin_normalizer import DEFAULT_MAPPING_PATH, normalize_simfin
 
@@ -61,13 +62,7 @@ def resolve_normalize_tickers(
 
 
 @click.command(context_settings={"help_option_names": ["-h", "--help"]})
-@click.option(
-    "--data-dir",
-    type=click.Path(path_type=Path, file_okay=False),
-    default=Path("data"),
-    show_default=True,
-    help="Data lake root.",
-)
+@lake_root_options
 @click.option(
     "--snapshot-date",
     type=click.DateTime(formats=["%Y-%m-%d"]),
@@ -105,7 +100,8 @@ def resolve_normalize_tickers(
     help="Suppress progress bars.",
 )
 def main(
-    data_dir: Path,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
     snapshot_date: datetime,
     universe_run_date: datetime | None,
     mapping: Path,
@@ -114,15 +110,20 @@ def main(
     quiet: bool,
 ) -> None:
     """Normalize raw SimFin bulk snapshots into curated PIT fundamentals."""
+    try:
+        lake_path = resolve_cli_data_dir(lake_root_uri=lake_root_uri, data_dir=data_dir)
+    except ValueError as exc:
+        raise click.ClickException(str(exc)) from exc
+
     snapshot = snapshot_date.date()
     universe_date = universe_run_date.date() if universe_run_date is not None else None
     ticker_set = resolve_normalize_tickers(
-        data_dir,
+        lake_path,
         universe_run_date=universe_date,
         explicit_tickers=tickers,
     )
     result = normalize_simfin(
-        data_dir,
+        lake_path,
         snapshot_date=snapshot,
         mapping_path=mapping,
         tickers=ticker_set,
diff --git a/src/smartwealthai/run_demo_pipeline.py b/src/smartwealthai/run_demo_pipeline.py
index a0772c9..62cac15 100644
--- a/src/smartwealthai/run_demo_pipeline.py
+++ b/src/smartwealthai/run_demo_pipeline.py
@@ -11,6 +11,7 @@
 import click
 from click.testing import CliRunner
 
+from smartwealthai.cli_lake import lake_root_options, resolve_cli_data_dir
 from smartwealthai.download_simfin import run_download
 from smartwealthai.magic_formula_ranking import ScoringResult, score_universe
 from smartwealthai.normalize_simfin import resolve_normalize_tickers
@@ -234,13 +235,7 @@ def format_pipeline_summary(result: PipelineResult) -> str:
 
 
 @click.command(context_settings={"help_option_names": ["-h", "--help"]})
-@click.option(
-    "--data-dir",
-    type=click.Path(path_type=Path, file_okay=False),
-    default=Path("data"),
-    show_default=True,
-    help="Data lake root.",
-)
+@lake_root_options
 @click.option(
     "--run-date",
     type=click.DateTime(formats=["%Y-%m-%d"]),
@@ -289,7 +284,8 @@ def format_pipeline_summary(result: PipelineResult) -> str:
     help="Skip SimFin bulk download (use existing raw lake).",
 )
 def main(
-    data_dir: Path,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
     run_date: datetime,
     snapshot_date: datetime | None,
     tickers: tuple[str, ...],
@@ -300,13 +296,18 @@ def main(
     skip_download: bool,
 ) -> None:
     """Run the demo slice: SimFin ingest, universe, normalize, prices, score-universe."""
+    try:
+        lake_path = resolve_cli_data_dir(lake_root_uri=lake_root_uri, data_dir=data_dir)
+    except ValueError as exc:
+        raise click.ClickException(str(exc)) from exc
+
     logging.basicConfig(level=logging.INFO, format="%(levelname)s %(message)s")
     decision_date = run_date.date()
     snapshot = snapshot_date.date() if snapshot_date is not None else decision_date
 
     try:
         result = run_demo_pipeline(
-            data_dir,
+            lake_path,
             run_date=decision_date,
             snapshot_date=snapshot,
             tickers=tickers,
diff --git a/src/smartwealthai/score_universe.py b/src/smartwealthai/score_universe.py
index 7551dc8..fb556a7 100644
--- a/src/smartwealthai/score_universe.py
+++ b/src/smartwealthai/score_universe.py
@@ -8,6 +8,7 @@
 import click
 from click.testing import CliRunner
 
+from smartwealthai.cli_lake import lake_root_options, resolve_cli_data_dir
 from smartwealthai.magic_formula_ranking import score_universe
 from smartwealthai.normalize_simfin import resolve_show_progress
 
@@ -40,13 +41,7 @@ def format_scoring_summary(result: object) -> str:
 
 
 @click.command(context_settings={"help_option_names": ["-h", "--help"]})
-@click.option(
-    "--data-dir",
-    type=click.Path(path_type=Path, file_okay=False),
-    default=Path("data"),
-    show_default=True,
-    help="Data lake root.",
-)
+@lake_root_options
 @click.option(
     "--run-date",
     type=click.DateTime(formats=["%Y-%m-%d"]),
@@ -72,17 +67,23 @@ def format_scoring_summary(result: object) -> str:
     help="Suppress progress bars.",
 )
 def main(
-    data_dir: Path,
+    lake_root_uri: str | None,
+    data_dir: Path | None,
     run_date: datetime,
     portfolio_size: int,
     progress: bool,
     quiet: bool,
 ) -> None:
     """Score the demo universe and write ranking + portfolio parquets."""
+    try:
+        lake_path = resolve_cli_data_dir(lake_root_uri=lake_root_uri, data_dir=data_dir)
+    except ValueError as exc:
+        raise click.ClickException(str(exc)) from exc
+
     decision_date = run_date.date()
     try:
         result = score_universe(
-            data_dir,
+            lake_path,
             run_date=decision_date,
             portfolio_size=portfolio_size,
             show_progress=resolve_show_progress(progress=progress, quiet=quiet),
diff --git a/tests/test_lake_root.py b/tests/test_lake_root.py
new file mode 100644
index 0000000..a764b9d
--- /dev/null
+++ b/tests/test_lake_root.py
@@ -0,0 +1,171 @@
+"""Hermetic tests for lake root URI resolution and file-backend I/O."""
+
+from __future__ import annotations
+
+import shutil
+from datetime import date
+from pathlib import Path
+
+import pandas as pd
+import pytest
+from click.testing import CliRunner
+
+from smartwealthai.cli_lake import resolve_cli_data_dir, resolve_cli_lake_root
+from smartwealthai.lake_paths import curated_portfolio_path
+from smartwealthai.lake_root import resolve_lake_root
+from smartwealthai.score_universe import cli_run as score_cli_run
+
+RUN_DATE = date(2026, 6, 18)
+FIXTURE_LAKE = Path("tests/fixtures/lake")
+
+
+def test_resolve_lake_root_file_uri_returns_usable_root(tmp_path: Path) -> None:
+    lake_dir = tmp_path / "lake"
+    lake_dir.mkdir()
+    root = resolve_lake_root(f"file://{lake_dir}")
+    expected = curated_portfolio_path(lake_dir, run_date=RUN_DATE)
+    assert root.curated_portfolio_path(run_date=RUN_DATE) == expected
+    assert root.as_path() == lake_dir.resolve()
+
+
+def test_resolve_lake_root_bare_path_equivalent_to_file_uri(tmp_path: Path) -> None:
+    lake_dir = tmp_path / "lake"
+    lake_dir.mkdir()
+    from_uri = resolve_lake_root(f"file://{lake_dir}")
+    from_bare = resolve_lake_root(str(lake_dir))
+    assert from_uri.as_path() == from_bare.as_path()
+
+
+def test_lake_root_parquet_round_trip(tmp_path: Path) -> None:
+    root = resolve_lake_root(str(tmp_path / "lake"))
+    path = root.curated_portfolio_path(run_date=RUN_DATE)
+    frame = pd.DataFrame({"ticker": ["AAPL"], "weight": [1.0]})
+    root.write_parquet(path, frame)
+    loaded = root.read_parquet(path)
+    pd.testing.assert_frame_equal(loaded, frame)
+
+
+def test_resolve_lake_root_rejects_unsupported_scheme() -> None:
+    with pytest.raises(ValueError, match="Unsupported lake root URI scheme"):
+        resolve_lake_root("ftp://example.com/lake")
+
+
+def test_resolve_lake_root_rejects_empty_uri() -> None:
+    with pytest.raises(ValueError, match="Lake root URI is required"):
+        resolve_lake_root(None)
+    with pytest.raises(ValueError, match="Lake root URI is required"):
+        resolve_lake_root("   ")
+
+
+def test_resolve_lake_root_rejects_invalid_s3_uri() -> None:
+    with pytest.raises(ValueError, match="Invalid S3 lake root URI"):
+        resolve_lake_root("s3://")
+
+
+def test_resolve_cli_lake_root_precedence(monkeypatch: pytest.MonkeyPatch, tmp_path: Path) -> None:
+    flag_path = tmp_path / "from-flag"
+    data_dir_path = tmp_path / "from-data-dir"
+    swai_path = tmp_path / "from-swai"
+    for path in (flag_path, data_dir_path, swai_path):
+        path.mkdir()
+
+    monkeypatch.setenv("SMARTWEALTHAI_DATA_DIR", str(swai_path))
+
+    flag_root = resolve_cli_lake_root(
+        lake_root_uri=f"file://{flag_path}",
+        data_dir=data_dir_path,
+    )
+    assert flag_root.as_path() == flag_path.resolve()
+
+    data_root = resolve_cli_lake_root(lake_root_uri=None, data_dir=data_dir_path)
+    assert data_root.as_path() == data_dir_path.resolve()
+
+    env_root = resolve_cli_lake_root(lake_root_uri=None, data_dir=None)
+    assert env_root.as_path() == swai_path.resolve()
+
+    monkeypatch.delenv("SMARTWEALTHAI_DATA_DIR", raising=False)
+    default_root = resolve_cli_lake_root(lake_root_uri=None, data_dir=None)
+    assert default_root.as_path() == Path("data").resolve()
+
+
+def test_resolve_cli_data_dir_returns_path(tmp_path: Path) -> None:
+    lake_dir = tmp_path / "lake"
+    lake_dir.mkdir()
+    path = resolve_cli_data_dir(lake_root_uri=f"file://{lake_dir}", data_dir=None)
+    assert path == lake_dir.resolve()
+
+
+@pytest.fixture
+def scored_lake(tmp_path: Path) -> Path:
+    """Minimal prepared lake for score-universe CLI regression."""
+    from smartwealthai.normalize_simfin import cli_run as normalize_cli_run
+    from smartwealthai.price_ingest import run_price_ingest
+    from smartwealthai.universe_builder import build_universe
+
+    shutil.copytree(FIXTURE_LAKE, tmp_path / "lake")
+    root = tmp_path / "lake"
+    build_universe(root, run_date=RUN_DATE, snapshot_date=RUN_DATE)
+    normalize_cli_run(
+        [
+            "--data-dir",
+            str(root),
+            "--snapshot-date",
+            RUN_DATE.isoformat(),
+            "--universe-run-date",
+            RUN_DATE.isoformat(),
+        ]
+    )
+    run_price_ingest(data_dir=root, run_date=RUN_DATE, snapshot_date=RUN_DATE)
+    return root
+
+
+def test_cli_score_universe_accepts_lake_root_uri_flag(scored_lake: Path) -> None:
+    exit_code = score_cli_run(
+        [
+            "--lake-root-uri",
+            f"file://{scored_lake}",
+            "--run-date",
+            RUN_DATE.isoformat(),
+            "--portfolio-size",
+            "3",
+        ]
+    )
+    assert exit_code == 0
+
+
+def test_cli_score_universe_data_dir_still_works(scored_lake: Path) -> None:
+    exit_code = score_cli_run(
+        [
+            "--data-dir",
+            str(scored_lake),
+            "--run-date",
+            RUN_DATE.isoformat(),
+            "--portfolio-size",
+            "3",
+        ]
+    )
+    assert exit_code == 0
+
+
+def test_cli_score_universe_rejects_malformed_lake_root_uri(scored_lake: Path) -> None:
+    runner = CliRunner()
+    result = runner.invoke(
+        __import__("smartwealthai.score_universe", fromlist=["main"]).main,
+        [
+            "--lake-root-uri",
+            "ftp://bad",
+            "--run-date",
+            RUN_DATE.isoformat(),
+        ],
+    )
+    assert result.exit_code != 0
+    assert "Unsupported lake root URI scheme" in result.output
+
+
+def test_s3_lake_root_parses_without_local_path() -> None:
+    root = resolve_lake_root("s3://smartwealthai-dev-lake/data/")
+    assert root.backend == "s3"
+    assert root.s3_bucket == "smartwealthai-dev-lake"
+    assert root.s3_prefix == "data/"
+    with pytest.raises(TypeError, match="local Path"):
+        root.as_path()
diff --git a/tests/test_quantitative_value_spec.py b/tests/test_quantitative_value_spec.py
new file mode 100644
index 0000000..41ff75c
--- /dev/null
+++ b/tests/test_quantitative_value_spec.py
@@ -0,0 +1,22 @@
+"""Tracer bullet: quantitative-value.md spec structure (issue #86)."""
+
+from pathlib import Path
+
+SPEC_PATH = Path("docs/mvp/features/quantitative-value.md")
+
+REQUIRED_HEADINGS = (
+    "## Objective",
+    "## MVP scope",
+    "## Acceptance criteria",
+)
+
+
+def test_quantitative_value_spec_exists_with_required_sections() -> None:
+  assert SPEC_PATH.is_file(), f"missing spec: {SPEC_PATH}"
+
+  text = SPEC_PATH.read_text(encoding="utf-8")
+
+  for heading in REQUIRED_HEADINGS:
+    assert heading in text, f"missing heading: {heading}"
+
+  assert "```mermaid" in text, "spec must include at least one Mermaid diagram"