From a43157cd44fdefd77cd8a959310be2e4cde651c8 Mon Sep 17 00:00:00 2001 From: Adriano Dal Pastro Date: Fri, 15 May 2026 19:23:12 +0000 Subject: [PATCH] refactor(gui): split dashboard in core (GA) + strategy_crypto (paper) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - NEW src/multi_swarm_core/multi_swarm_core/dashboard/ con theme.py + data.py + nicegui_app.py - theme.py condiviso (CSS + colors + _apply_theme + _json_to_html + _build_header parametrico) - core GUI: pagine /, /convergence, /genomes — legge SOLO runs.db - strategy GUI slim: solo /, legge SOLO strategy_crypto.db — importa theme dal core - Aggiunto nicegui+plotly al core pyproject (uv.lock rigenerato) - docker-compose: nuovo servizio multi-swarm-core-gui su /multi_swarm_core_gui (Traefik PathPrefix + replacepathregex, NO stripprefix per evitare doppio root_path) - .env.example: DASHBOARD_ROOT_PATH ora per-servizio Pattern: ogni modulo possiede la sua GUI, ogni GUI legge solo il proprio DB. N strategie future = duplica lo scheletro strategy_crypto/frontend/. Co-Authored-By: Claude Opus 4.7 (1M context) --- .env.example | 9 +- docker-compose.yml | 55 +- .../multi_swarm_core/dashboard/__init__.py | 0 .../multi_swarm_core/dashboard/data.py | 68 ++ .../multi_swarm_core/dashboard/nicegui_app.py | 560 ++++++++++++ .../multi_swarm_core/dashboard/theme.py | 383 ++++++++ src/multi_swarm_core/pyproject.toml | 2 + .../strategy_crypto/frontend/data.py | 52 +- .../strategy_crypto/frontend/nicegui_app.py | 863 +----------------- uv.lock | 4 + 10 files changed, 1092 insertions(+), 904 deletions(-) create mode 100644 src/multi_swarm_core/multi_swarm_core/dashboard/__init__.py create mode 100644 src/multi_swarm_core/multi_swarm_core/dashboard/data.py create mode 100644 src/multi_swarm_core/multi_swarm_core/dashboard/nicegui_app.py create mode 100644 src/multi_swarm_core/multi_swarm_core/dashboard/theme.py diff --git a/.env.example b/.env.example index bf1dc0e..3cf1a35 100644 --- a/.env.example +++ b/.env.example @@ -29,12 +29,15 @@ GA_DB_PATH=./state/runs.db STRATEGY_CRYPTO_DB_PATH=./state/strategy_crypto.db # Docker / Traefik (usati SOLO da docker-compose.yml) -# Dominio base: traefik espone la dashboard su swarm.${DOMAIN_NAME}/strategy_crypto_gui +# Dominio base: traefik espone le dashboard su swarm.${DOMAIN_NAME}/... DOMAIN_NAME=tielogic.xyz # Porta interna della NiceGUI dashboard (Traefik fa il TLS davanti) SWARM_DASHBOARD_PORT=8080 -# Subpath URL del dashboard NiceGUI (usato come root_path in produzione) -DASHBOARD_ROOT_PATH=/strategy_crypto_gui +# Subpath URL del dashboard NiceGUI — ora PER-SERVIZIO nel docker-compose.yml: +# strategy-crypto-gui -> DASHBOARD_ROOT_PATH=/strategy_crypto_gui +# multi-swarm-core-gui -> DASHBOARD_ROOT_PATH=/multi_swarm_core_gui +# In sviluppo locale lascia vuoto (nessun subpath). +DASHBOARD_ROOT_PATH= # Paper-trading runner — override del command nel compose (opzionali) PAPER_RUN_NAME=phase3-papertrade-prod diff --git a/docker-compose.yml b/docker-compose.yml index ee4dd42..580611e 100644 --- a/docker-compose.yml +++ b/docker-compose.yml @@ -1,12 +1,14 @@ # docker-compose.yml — Multi-Swarm Coevolutive # -# Due servizi della strategia crypto, condividono la stessa immagine +# Tre servizi della strategia crypto, condividono la stessa immagine # `multi-swarm-coevolutive:dev` buildata dal Dockerfile root (uv workspace): # -# * strategy-crypto-paper — paper-trading runner long-running -# (scripts/run_paper_trading.py) -# * strategy-crypto-gui — NiceGUI dashboard esposta da Traefik su -# https://swarm.${DOMAIN_NAME:-tielogic.xyz}/strategy_crypto_gui +# * strategy-crypto-paper — paper-trading runner long-running +# (scripts/run_paper_trading.py) +# * strategy-crypto-gui — NiceGUI dashboard esposta da Traefik su +# https://swarm.${DOMAIN_NAME:-tielogic.xyz}/strategy_crypto_gui +# * multi-swarm-core-gui — NiceGUI dashboard GA esposta su +# https://swarm.${DOMAIN_NAME:-tielogic.xyz}/multi_swarm_core_gui # # Entrambi joinano la rete external `traefik` cosi' il client Cerbero # risolve direttamente l'host `cerbero-mcp` (porta 9000) senza passare @@ -36,12 +38,11 @@ x-swarm-env: &swarm-env # DB separati per dominio: GA_DB_PATH: /app/state/runs.db STRATEGY_CRYPTO_DB_PATH: /app/state/strategy_crypto.db - # Subpath sotto cui la dashboard NiceGUI e' esposta da Traefik. + # DASHBOARD_ROOT_PATH e' ora per-servizio (vedi environment blocks sotto). # IMPORTANT: NON usare StripPrefix middleware con questo. NiceGUI/Starlette # gestisce internamente il root_path su request path che ARRIVANO con prefix. # StripPrefix causa doppio prefix negli asset URL (NiceGUI prefixa + uvicorn # rilegge X-Forwarded-Prefix e prefixa di nuovo). - DASHBOARD_ROOT_PATH: /strategy_crypto_gui services: strategy-crypto-paper: @@ -85,11 +86,10 @@ services: env_file: .env environment: <<: *swarm-env + DASHBOARD_ROOT_PATH: /strategy_crypto_gui volumes: - # Dashboard legge entrambi i DB: state/ in read-only (WAL: vedi nota) + # Dashboard legge solo strategy_crypto.db: state/ in read-only (WAL: vedi nota) - ./state:/app/state:ro - - ./data:/app/data:ro - - ./series:/app/series:ro entrypoint: - python - -m @@ -121,3 +121,38 @@ services: - "traefik.http.middlewares.strategy-crypto-replace.replacepathregex.replacement=$$1" - "traefik.http.routers.strategy-crypto-gui.middlewares=strategy-crypto-replace" - com.centurylinklabs.watchtower.enable=true + + multi-swarm-core-gui: + image: multi-swarm-coevolutive:dev + build: + context: . + dockerfile: Dockerfile + container_name: multi-swarm-core-gui + restart: unless-stopped + networks: [traefik] + env_file: .env + environment: + <<: *swarm-env + DASHBOARD_ROOT_PATH: /multi_swarm_core_gui + volumes: + - ./state:/app/state:ro + entrypoint: [python, -m, multi_swarm_core.dashboard.nicegui_app] + command: [] + healthcheck: + test: ["CMD", "python", "-c", "import os, urllib.request; urllib.request.urlopen(f'http://localhost:{os.environ.get(\"SWARM_DASHBOARD_PORT\",\"8080\")}/', timeout=3).close()"] + interval: 30s + timeout: 5s + retries: 3 + start_period: 30s + labels: + - traefik.enable=true + - traefik.docker.network=traefik + - "traefik.http.routers.multi-swarm-core-gui.rule=Host(`swarm.${DOMAIN_NAME:-tielogic.xyz}`) && PathPrefix(`/multi_swarm_core_gui`)" + - traefik.http.routers.multi-swarm-core-gui.tls=true + - traefik.http.routers.multi-swarm-core-gui.entrypoints=websecure + - traefik.http.routers.multi-swarm-core-gui.tls.certresolver=mytlschallenge + - "traefik.http.services.multi-swarm-core-gui.loadbalancer.server.port=${SWARM_DASHBOARD_PORT:-8080}" + - "traefik.http.middlewares.multi-swarm-core-replace.replacepathregex.regex=^/multi_swarm_core_gui(/.*|$$)" + - "traefik.http.middlewares.multi-swarm-core-replace.replacepathregex.replacement=$$1" + - "traefik.http.routers.multi-swarm-core-gui.middlewares=multi-swarm-core-replace" + - com.centurylinklabs.watchtower.enable=true diff --git a/src/multi_swarm_core/multi_swarm_core/dashboard/__init__.py b/src/multi_swarm_core/multi_swarm_core/dashboard/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/src/multi_swarm_core/multi_swarm_core/dashboard/data.py b/src/multi_swarm_core/multi_swarm_core/dashboard/data.py new file mode 100644 index 0000000..5829118 --- /dev/null +++ b/src/multi_swarm_core/multi_swarm_core/dashboard/data.py @@ -0,0 +1,68 @@ +"""GA data access functions for the core dashboard. + +Reads exclusively from runs.db (GA tables). +""" + +from __future__ import annotations + +import json +from pathlib import Path +from typing import Any + +import pandas as pd # type: ignore[import-untyped] + +from multi_swarm_core.persistence.repository import Repository + +__all__ = [ + "get_repo", + "list_runs_df", + "get_run_overview", + "generations_df", + "evaluations_df", + "genomes_df", +] + + +def get_repo(db_path: str | Path) -> Repository: + return Repository(db_path=db_path) + + +def list_runs_df(repo: Repository) -> pd.DataFrame: + return pd.DataFrame(repo.list_runs()) + + +def get_run_overview(repo: Repository, run_id: str) -> dict[str, Any]: + run = repo.get_run(run_id) + return { + "name": run["name"], + "started_at": run["started_at"], + "completed_at": run["completed_at"], + "status": run["status"], + "total_cost_usd": run["total_cost_usd"], + "config": json.loads(run["config_json"]), + } + + +def generations_df(repo: Repository, run_id: str) -> pd.DataFrame: + return pd.DataFrame(repo.list_generations(run_id)) + + +def evaluations_df(repo: Repository, run_id: str) -> pd.DataFrame: + return pd.DataFrame(repo.list_evaluations(run_id)) + + +def genomes_df( + repo: Repository, run_id: str, generation_idx: int | None = None +) -> pd.DataFrame: + rows = repo.list_genomes(run_id, generation_idx) + flat: list[dict[str, Any]] = [] + for r in rows: + payload = json.loads(r["payload_json"]) + flat.append( + { + "id": r["id"], + "generation_idx": r["generation_idx"], + **payload, + } + ) + return pd.DataFrame(flat) diff --git a/src/multi_swarm_core/multi_swarm_core/dashboard/nicegui_app.py b/src/multi_swarm_core/multi_swarm_core/dashboard/nicegui_app.py new file mode 100644 index 0000000..4ccfc38 --- /dev/null +++ b/src/multi_swarm_core/multi_swarm_core/dashboard/nicegui_app.py @@ -0,0 +1,560 @@ +"""Multi-Swarm Core Dashboard — GA pages: /, /convergence, /genomes. + +Avvio: ``uv run python -m multi_swarm_core.dashboard.nicegui_app`` +Legge SOLO runs.db (tabelle GA). +""" + +from __future__ import annotations + +import html +import os +from pathlib import Path +from typing import Any + +import plotly.graph_objects as go # type: ignore[import-untyped] +from nicegui import app, ui + +from multi_swarm_core.dashboard.data import ( + evaluations_df, + generations_df, + genomes_df, + get_repo, + get_run_overview, + list_runs_df, +) +from multi_swarm_core.dashboard.theme import ( + COLOR_ACCENT, + COLOR_PRIMARY, + COLOR_SECONDARY, + COLOR_SURFACE, + COLOR_TEXT, + COLOR_TEXT_MUTED, + _STATUS_BADGE, + _apply_theme, + _build_header, + _json_to_html, +) + +GA_DB_PATH = os.environ.get("GA_DB_PATH", "./state/runs.db") +DASHBOARD_ROOT_PATH = os.environ.get("DASHBOARD_ROOT_PATH", "") +REFRESH_INTERVAL_S = 3.0 + + +def _runs_options() -> dict[str, str]: + repo = get_repo(GA_DB_PATH) + runs = list_runs_df(repo) + if runs.empty: + return {} + return { + row["id"]: f"{row['name']} — {row['status']} ({row['started_at'][:16]})" + for _, row in runs.iterrows() + } + + +def _snapshot(run_id: str) -> dict[str, Any]: + repo = get_repo(GA_DB_PATH) + ov = get_run_overview(repo, run_id) + evals = evaluations_df(repo, run_id) + gens = generations_df(repo, run_id) + + cfg = ov["config"] + pop_size = int(cfg.get("population_size", 0)) + n_gens = int(cfg.get("n_generations", 0)) + evals_total = max(pop_size * n_gens, 1) + evals_done = len(evals) + gens_done = int(gens["completed_at"].notna().sum()) if not gens.empty else 0 + live_cost = float(repo.total_cost(run_id)) if ov["status"] == "running" else float( + ov["total_cost_usd"] + ) + + top_fit = float(evals["fitness"].max()) if evals_done else float("nan") + median_fit = float(evals["fitness"].median()) if evals_done else float("nan") + parse_success = ( + 100.0 * float(evals["parse_error"].isna().sum()) / evals_done if evals_done else 0.0 + ) + + return { + "status": ov["status"], + "name": cfg.get("run_name", "—"), + "started_at": ov["started_at"], + "completed_at": ov["completed_at"] or "—", + "cost_usd": live_cost, + "pop_size": pop_size, + "n_gens": n_gens, + "evals_done": evals_done, + "evals_total": evals_total, + "gens_done": gens_done, + "top_fit": top_fit, + "median_fit": median_fit, + "parse_success": parse_success, + "config": cfg, + "gens_df": gens, + } + + +def _convergence_figure(gens_df: Any) -> go.Figure: + fig = go.Figure() + if gens_df.empty: + fig.add_annotation( + text="Nessuna generazione registrata", x=0.5, y=0.5, showarrow=False, + font={"color": COLOR_TEXT_MUTED, "size": 14}, + ) + else: + fig.add_trace( + go.Scatter( + x=gens_df["generation_idx"], y=gens_df["fitness_max"], + name="max", mode="lines+markers", + line={"color": COLOR_PRIMARY, "width": 3, "shape": "spline", "smoothing": 0.6}, + marker={"size": 9, "color": COLOR_PRIMARY, + "line": {"color": "#fff", "width": 1}}, + fill="tozeroy", + fillcolor="rgba(255, 45, 135, 0.12)", + ) + ) + fig.add_trace( + go.Scatter( + x=gens_df["generation_idx"], y=gens_df["fitness_p90"], + name="p90", mode="lines+markers", + line={"color": COLOR_ACCENT, "width": 2, "dash": "dot", "shape": "spline"}, + marker={"size": 7, "color": COLOR_ACCENT}, + ) + ) + fig.add_trace( + go.Scatter( + x=gens_df["generation_idx"], y=gens_df["fitness_median"], + name="median", mode="lines+markers", + line={"color": COLOR_SECONDARY, "width": 2, "shape": "spline"}, + marker={"size": 7, "color": COLOR_SECONDARY}, + ) + ) + fig.update_layout( + template="plotly_dark", + paper_bgcolor=COLOR_SURFACE, + plot_bgcolor=COLOR_SURFACE, + font={"color": COLOR_TEXT}, + xaxis={"title": "generation", "gridcolor": "rgba(148, 163, 184, 0.08)", "dtick": 1}, + yaxis={"title": "fitness", "gridcolor": "rgba(148, 163, 184, 0.08)"}, + title={"text": "Fitness convergence", "font": {"color": COLOR_TEXT, "size": 18}}, + legend={"bgcolor": "rgba(19, 19, 26, 0.95)", "bordercolor": COLOR_PRIMARY, "borderwidth": 1}, + margin={"l": 50, "r": 30, "t": 50, "b": 50}, + ) + return fig + + +def _entropy_figure(gens_df: Any) -> go.Figure: + fig = go.Figure() + if not gens_df.empty: + fig.add_trace( + go.Scatter( + x=gens_df["generation_idx"], y=gens_df["entropy"], + mode="lines+markers", + line={"color": COLOR_SECONDARY, "width": 3, "shape": "spline", "smoothing": 0.6}, + marker={"size": 9, "color": COLOR_SECONDARY, + "line": {"color": "#fff", "width": 1}}, + fill="tozeroy", + fillcolor="rgba(0, 217, 255, 0.12)", + name="entropy", + ) + ) + fig.add_hline( + y=0.5, line_dash="dash", line_color=COLOR_ACCENT, + annotation_text="gate threshold (0.5)", + annotation_font_color=COLOR_ACCENT, + ) + fig.update_layout( + template="plotly_dark", + paper_bgcolor=COLOR_SURFACE, + plot_bgcolor=COLOR_SURFACE, + font={"color": COLOR_TEXT}, + xaxis={"title": "generation", "gridcolor": "rgba(148, 163, 184, 0.08)", "dtick": 1}, + yaxis={"title": "entropy", "gridcolor": "rgba(148, 163, 184, 0.08)"}, + title={"text": "Diversity (fitness entropy)", "font": {"color": COLOR_TEXT, "size": 18}}, + margin={"l": 50, "r": 30, "t": 50, "b": 50}, + ) + return fig + + +@ui.page("/") +def index() -> None: + _apply_theme() + _build_header( + active="/", + brand_subtitle="Coevolutivo / GA", + nav_items=[("/", "Overview"), ("/convergence", "Convergence"), ("/genomes", "Genomes")], + db_label=f"⛁ {Path(GA_DB_PATH).resolve().name}", + ) + + options = _runs_options() + if not options: + ui.label("Nessuna run nel database.").classes("text-h5") + return + + state: dict[str, Any] = {"run_id": next(iter(options))} + + with ui.row().classes("w-full items-center gap-4 q-mb-md"): + select = ui.select(options=options, value=state["run_id"], label="Run").classes( + "flex-grow" + ) + status_badge = ui.badge("…", color="primary").classes("text-body1 q-pa-sm") + ui.button("🔄 Refresh", on_click=lambda: refresh()).props("outline color=primary") + + with ui.card().classes("w-full"): + ui.label("Progresso run").classes("text-subtitle1") + gen_label = ui.label("Generations: 0/0") + gen_bar = ui.linear_progress(0.0, show_value=False).props("size=20px color=primary") + eval_label = ui.label("Evaluations: 0/0 (0.0%)") + eval_bar = ui.linear_progress(0.0, show_value=False).props("size=20px color=accent") + + with ui.row().classes("w-full gap-4"): + with ui.card().classes("flex-grow metric-card accent-cyan"): + ui.label("Top fitness").classes("text-caption") + top_lbl = ui.label("—").classes("text-h4") + with ui.card().classes("flex-grow metric-card accent-purple"): + ui.label("Median fitness").classes("text-caption") + median_lbl = ui.label("—").classes("text-h4") + with ui.card().classes("flex-grow metric-card accent-amber"): + ui.label("Parse success").classes("text-caption") + parse_lbl = ui.label("—").classes("text-h4") + with ui.card().classes("flex-grow metric-card accent-green"): + ui.label("Cost (USD)").classes("text-caption") + cost_lbl = ui.label("—").classes("text-h4") + + with ui.row().classes("w-full gap-4 q-mt-md"): + started_lbl = ui.label("Started: —") + completed_lbl = ui.label("Completed: —") + ui.separator() + ui.label("Config").classes("text-subtitle1") + cfg_code = ui.html('
').classes("w-full")
+
+    def refresh() -> None:
+        run_id = select.value
+        if not run_id:
+            return
+        try:
+            s = _snapshot(run_id)
+        except Exception as e:  # noqa: BLE001
+            ui.notify(f"Errore: {e}", type="negative")
+            return
+
+        text, color = _STATUS_BADGE.get(s["status"], (s["status"], "primary"))
+        status_badge.text = text
+        status_badge.props(f"color={color}")
+
+        gen_frac = min(s["gens_done"] / max(s["n_gens"], 1), 1.0)
+        eval_frac = min(s["evals_done"] / s["evals_total"], 1.0)
+        gen_bar.value = gen_frac
+        eval_bar.value = eval_frac
+        gen_label.text = f"Generations: {s['gens_done']}/{s['n_gens']}"
+        eval_label.text = (
+            f"Evaluations: {s['evals_done']}/{s['evals_total']} ({100 * eval_frac:.1f}%)"
+        )
+
+        top_lbl.text = f"{s['top_fit']:.4f}" if s["evals_done"] else "—"
+        median_lbl.text = f"{s['median_fit']:.4f}" if s["evals_done"] else "—"
+        parse_lbl.text = f"{s['parse_success']:.1f}%" if s["evals_done"] else "—"
+        cost_lbl.text = f"${s['cost_usd']:.4f}"
+
+        started_lbl.text = f"Started: {s['started_at']}"
+        completed_lbl.text = f"Completed: {s['completed_at']}"
+        cfg_code.content = f'
{_json_to_html(s["config"])}
'
+
+    def on_select_change() -> None:
+        state["run_id"] = select.value
+        refresh()
+
+    select.on_value_change(on_select_change)
+    ui.timer(REFRESH_INTERVAL_S, refresh)
+    refresh()
+
+
+@ui.page("/convergence")
+def convergence() -> None:
+    _apply_theme()
+    _build_header(
+        active="/convergence",
+        brand_subtitle="Coevolutivo / GA",
+        nav_items=[("/", "Overview"), ("/convergence", "Convergence"), ("/genomes", "Genomes")],
+        db_label=f"⛁ {Path(GA_DB_PATH).resolve().name}",
+    )
+
+    options = _runs_options()
+    if not options:
+        ui.label("Nessuna run nel database.").classes("text-h5")
+        return
+
+    state: dict[str, Any] = {"run_id": next(iter(options))}
+
+    with ui.row().classes("w-full items-center gap-4 q-mb-md"):
+        select = ui.select(options=options, value=state["run_id"], label="Run").classes(
+            "flex-grow"
+        )
+        gen_count_lbl = ui.label("Gens: 0/0").classes("text-body1").style(
+            f"color: {COLOR_PRIMARY}; font-weight: 600;"
+        )
+        ui.button("🔄 Refresh", on_click=lambda: refresh()).props("outline color=primary")
+
+    fitness_plot = ui.plotly(_convergence_figure(generations_df(get_repo(GA_DB_PATH), state["run_id"]))).classes("w-full")
+    entropy_plot = ui.plotly(_entropy_figure(generations_df(get_repo(GA_DB_PATH), state["run_id"]))).classes("w-full q-mt-md")
+
+    ui.separator()
+    ui.label("Tabella generazioni").classes("text-subtitle1 q-mt-md")
+    gens_table = ui.table(
+        columns=[
+            {"name": "generation_idx", "label": "gen", "field": "generation_idx", "sortable": True},
+            {"name": "n_genomes", "label": "n", "field": "n_genomes"},
+            {"name": "fitness_max", "label": "max", "field": "fitness_max"},
+            {"name": "fitness_p90", "label": "p90", "field": "fitness_p90"},
+            {"name": "fitness_median", "label": "median", "field": "fitness_median"},
+            {"name": "entropy", "label": "entropy", "field": "entropy"},
+            {"name": "completed_at", "label": "completed", "field": "completed_at"},
+        ],
+        rows=[],
+        row_key="generation_idx",
+    ).classes("w-full")
+
+    def refresh() -> None:
+        run_id = select.value
+        if not run_id:
+            return
+        try:
+            gens = generations_df(get_repo(GA_DB_PATH), run_id)
+            ov = get_run_overview(get_repo(GA_DB_PATH), run_id)
+        except Exception as e:  # noqa: BLE001
+            ui.notify(f"Errore: {e}", type="negative")
+            return
+
+        n_gens = int(ov["config"].get("n_generations", 0))
+        gens_done = int(gens["completed_at"].notna().sum()) if not gens.empty else 0
+        gen_count_lbl.text = f"Gens: {gens_done}/{n_gens}"
+
+        fitness_plot.update_figure(_convergence_figure(gens))
+        entropy_plot.update_figure(_entropy_figure(gens))
+
+        if gens.empty:
+            gens_table.rows = []
+        else:
+            display_cols = [
+                "generation_idx", "n_genomes",
+                "fitness_max", "fitness_p90", "fitness_median",
+                "entropy", "completed_at",
+            ]
+            gens_table.rows = [
+                {
+                    col: (round(v, 6) if isinstance(v, float) else v)
+                    for col, v in row.items()
+                    if col in display_cols
+                }
+                for _, row in gens.iterrows()
+            ]
+        gens_table.update()
+
+    def on_select_change() -> None:
+        state["run_id"] = select.value
+        refresh()
+
+    select.on_value_change(on_select_change)
+    ui.timer(REFRESH_INTERVAL_S, refresh)
+    refresh()
+
+
+@ui.page("/genomes")
+def genomes() -> None:
+    _apply_theme()
+    _build_header(
+        active="/genomes",
+        brand_subtitle="Coevolutivo / GA",
+        nav_items=[("/", "Overview"), ("/convergence", "Convergence"), ("/genomes", "Genomes")],
+        db_label=f"⛁ {Path(GA_DB_PATH).resolve().name}",
+    )
+
+    options = _runs_options()
+    if not options:
+        ui.label("Nessuna run nel database.").classes("text-h5")
+        return
+
+    state: dict[str, Any] = {
+        "run_id": next(iter(options)),
+        "selected_gid": None,
+        "merged": None,
+    }
+
+    with ui.row().classes("w-full items-center gap-4 q-mb-md"):
+        select = ui.select(options=options, value=state["run_id"], label="Run").classes(
+            "flex-grow"
+        )
+        top_k_select = ui.select(
+            options={10: "Top 10", 25: "Top 25", 50: "Top 50"},
+            value=10,
+            label="Top K",
+        )
+        ui.button("🔄 Refresh", on_click=lambda: refresh()).props("outline color=primary")
+
+    ui.label("Top genomi per fitness").classes("text-subtitle1 q-mt-sm")
+    top_table = ui.table(
+        columns=[
+            {"name": "genome_id", "label": "id", "field": "genome_id", "align": "left"},
+            {"name": "fitness", "label": "fitness", "field": "fitness", "sortable": True},
+            {"name": "dsr", "label": "DSR", "field": "dsr"},
+            {"name": "sharpe", "label": "Sharpe", "field": "sharpe"},
+            {"name": "max_dd", "label": "max DD", "field": "max_dd"},
+            {"name": "n_trades", "label": "trades", "field": "n_trades"},
+            {"name": "cognitive_style", "label": "style", "field": "cognitive_style"},
+            {"name": "temperature", "label": "T", "field": "temperature"},
+            {"name": "lookback_window", "label": "lookback", "field": "lookback_window"},
+        ],
+        rows=[],
+        row_key="genome_id",
+        selection="single",
+    ).classes("w-full")
+
+    ui.separator().classes("q-my-md")
+
+    with ui.card().classes("w-full"):
+        ui.label("Ispezione genoma").classes("text-subtitle1")
+        detail_hint = ui.label("Seleziona un genoma dalla tabella sopra.").classes(
+            "text-caption"
+        ).style(f"color: {COLOR_TEXT_MUTED};")
+
+        with ui.row().classes("w-full gap-4 q-mt-sm"):
+            with ui.card().classes("flex-grow metric-card accent-cyan"):
+                ui.label("fitness").classes("text-caption")
+                fit_lbl = ui.label("—").classes("text-h4")
+            with ui.card().classes("flex-grow metric-card accent-purple"):
+                ui.label("DSR").classes("text-caption")
+                dsr_lbl = ui.label("—").classes("text-h4")
+            with ui.card().classes("flex-grow metric-card accent-amber"):
+                ui.label("Sharpe").classes("text-caption")
+                sharpe_lbl = ui.label("—").classes("text-h4")
+            with ui.card().classes("flex-grow metric-card"):
+                ui.label("max DD").classes("text-caption")
+                dd_lbl = ui.label("—").classes("text-h4")
+            with ui.card().classes("flex-grow metric-card accent-green"):
+                ui.label("trades").classes("text-caption")
+                trades_lbl = ui.label("—").classes("text-h4")
+            with ui.card().classes("flex-grow metric-card"):
+                ui.label("style").classes("text-caption")
+                style_lbl = ui.label("—").classes("text-h4")
+
+        ui.label("System prompt").classes("text-subtitle1 q-mt-md")
+        prompt_code = ui.html('
—
').classes("w-full")
+
+        ui.label("Raw LLM output").classes("text-subtitle1 q-mt-md")
+        raw_code = ui.html('
—
').classes("w-full")
+
+        parse_error_lbl = ui.label("").classes("q-mt-sm").style(
+            "color: #FF6B6B; font-weight: 600;"
+        )
+
+    def _render_detail(row: dict[str, Any]) -> None:
+        detail_hint.text = f"Genoma: {row.get('genome_id', '—')}"
+        fit_lbl.text = f"{float(row.get('fitness', 0)):.4f}"
+        dsr_lbl.text = f"{float(row.get('dsr', 0)):.4f}"
+        sharpe_lbl.text = f"{float(row.get('sharpe', 0)):.3f}"
+        dd_lbl.text = f"{float(row.get('max_dd', 0)):.3f}"
+        trades_lbl.text = str(int(row.get("n_trades", 0)))
+        style_lbl.text = str(row.get("cognitive_style", "—"))
+        prompt_code.content = (
+            f'
{html.escape(str(row.get("system_prompt", "—")))}
'
+        )
+        raw_code.content = (
+            f'
{html.escape(str(row.get("raw_text", "—") or "—"))}
'
+        )
+        pe = row.get("parse_error")
+        parse_error_lbl.text = f"❌ Parse error: {pe}" if pe else ""
+
+    def refresh() -> None:
+        run_id = select.value
+        if not run_id:
+            return
+        try:
+            repo = get_repo(GA_DB_PATH)
+            evals = evaluations_df(repo, run_id)
+            gens = genomes_df(repo, run_id)
+        except Exception as e:  # noqa: BLE001
+            ui.notify(f"Errore: {e}", type="negative")
+            return
+
+        if evals.empty:
+            top_table.rows = []
+            top_table.update()
+            return
+
+        merged = evals.merge(
+            gens, left_on="genome_id", right_on="id", how="left", suffixes=("", "_g")
+        )
+        state["merged"] = merged
+
+        k = int(top_k_select.value)
+        top = merged.sort_values("fitness", ascending=False).head(k)
+
+        rows = []
+        for _, r in top.iterrows():
+            rows.append(
+                {
+                    "genome_id": str(r.get("genome_id", "—"))[:12] + "…",
+                    "fitness": round(float(r.get("fitness", 0)), 4),
+                    "dsr": round(float(r.get("dsr", 0)), 4),
+                    "sharpe": round(float(r.get("sharpe", 0)), 3),
+                    "max_dd": round(float(r.get("max_dd", 0)), 3),
+                    "n_trades": int(r.get("n_trades", 0)),
+                    "cognitive_style": str(r.get("cognitive_style", "—")),
+                    "temperature": round(float(r.get("temperature", 0)), 2),
+                    "lookback_window": int(r.get("lookback_window", 0)),
+                    "_full_id": str(r.get("genome_id", "")),
+                }
+            )
+        top_table.rows = rows
+        top_table.update()
+
+        sel = state.get("selected_gid")
+        if sel:
+            match = merged[merged["genome_id"] == sel]
+            if not match.empty:
+                _render_detail(match.iloc[0].to_dict())
+
+    def on_row_selected(e: Any) -> None:
+        rows = (e.args or {}).get("rows") or []
+        if not rows:
+            return
+        full_id = rows[0].get("_full_id")
+        if not full_id:
+            return
+        state["selected_gid"] = full_id
+        merged = state.get("merged")
+        if merged is None:
+            return
+        match = merged[merged["genome_id"] == full_id]
+        if not match.empty:
+            _render_detail(match.iloc[0].to_dict())
+
+    def on_select_change() -> None:
+        state["run_id"] = select.value
+        state["selected_gid"] = None
+        refresh()
+
+    select.on_value_change(on_select_change)
+    top_k_select.on_value_change(lambda _: refresh())
+    top_table.on("selection", on_row_selected)
+    ui.timer(REFRESH_INTERVAL_S, refresh)
+    refresh()
+
+
+def main() -> None:
+    app.on_startup(
+        lambda: print(
+            f"GA DB: {Path(GA_DB_PATH).resolve()} | root_path: {DASHBOARD_ROOT_PATH or '/'}"
+        )
+    )
+    ui.run(
+        host="0.0.0.0",
+        port=int(os.environ.get("SWARM_DASHBOARD_PORT", "8080")),
+        title="Multi-Swarm Core Dashboard",
+        reload=False,
+        show=False,
+        dark=True,
+        root_path=DASHBOARD_ROOT_PATH,
+    )
+
+
+if __name__ in {"__main__", "__mp_main__"}:
+    main()
diff --git a/src/multi_swarm_core/multi_swarm_core/dashboard/theme.py b/src/multi_swarm_core/multi_swarm_core/dashboard/theme.py
new file mode 100644
index 0000000..20c0cec
--- /dev/null
+++ b/src/multi_swarm_core/multi_swarm_core/dashboard/theme.py
@@ -0,0 +1,383 @@
+"""Shared theme module for NiceGUI dashboards.
+
+Exports palette constants, CSS, and helper functions used by both
+multi_swarm_core.dashboard and strategy_crypto.frontend dashboards.
+"""
+
+from __future__ import annotations
+
+import html
+from typing import Any
+
+from nicegui import ui
+
+__all__ = [
+    "COLOR_BG",
+    "COLOR_SURFACE",
+    "COLOR_SURFACE_2",
+    "COLOR_BORDER",
+    "COLOR_BORDER_HOVER",
+    "COLOR_PRIMARY",
+    "COLOR_SECONDARY",
+    "COLOR_ACCENT",
+    "COLOR_SUCCESS",
+    "COLOR_DANGER",
+    "COLOR_TEXT",
+    "COLOR_TEXT_MUTED",
+    "_STATUS_BADGE",
+    "_CUSTOM_CSS",
+    "_json_to_html",
+    "_apply_theme",
+    "_build_header",
+]
+
+# --- Neon Trading Dashboard palette ---
+COLOR_BG = "#0A0A0F"
+COLOR_SURFACE = "#13131A"
+COLOR_SURFACE_2 = "#1C1C26"
+COLOR_BORDER = "rgba(255, 45, 135, 0.12)"
+COLOR_BORDER_HOVER = "rgba(255, 45, 135, 0.45)"
+COLOR_PRIMARY = "#FF2D87"
+COLOR_SECONDARY = "#00D9FF"
+COLOR_ACCENT = "#FFB800"
+COLOR_SUCCESS = "#00E676"
+COLOR_DANGER = "#FF3D60"
+COLOR_TEXT = "#FFFFFF"
+COLOR_TEXT_MUTED = "#7A7A8C"
+
+_STATUS_BADGE: dict[str, tuple[str, str]] = {
+    "running": ("● running", "positive"),
+    "completed": ("✓ completed", "positive"),
+    "failed": ("✕ failed", "negative"),
+}
+
+_CUSTOM_CSS = f"""
+
+"""
+
+
+def _json_to_html(obj: Any, indent: int = 0) -> str:
+    """Render JSON con span colorati espliciti. Garantisce leggibilità ovunque."""
+    pad = "  " * indent
+    inner_pad = "  " * (indent + 1)
+    if isinstance(obj, dict):
+        if not obj:
+            return '{}'
+        items = []
+        for k, v in obj.items():
+            key = f'"{html.escape(str(k))}"'
+            val = _json_to_html(v, indent + 1)
+            items.append(f"{inner_pad}{key}: {val}")
+        return ('{\n'
+                + ',\n'.join(items)
+                + f'\n{pad}}}')
+    if isinstance(obj, list):
+        if not obj:
+            return '[]'
+        items = [_json_to_html(x, indent + 1) for x in obj]
+        return ('[\n'
+                + ',\n'.join(inner_pad + i for i in items)
+                + f'\n{pad}]')
+    if isinstance(obj, bool):
+        return f'{str(obj).lower()}'
+    if obj is None:
+        return 'null'
+    if isinstance(obj, (int, float)):
+        return f'{obj}'
+    return f'"{html.escape(str(obj))}"'
+
+
+def _apply_theme() -> None:
+    ui.add_head_html(_CUSTOM_CSS)
+    ui.dark_mode().enable()
+    ui.colors(
+        primary=COLOR_PRIMARY,
+        secondary=COLOR_SECONDARY,
+        accent=COLOR_ACCENT,
+        dark=COLOR_BG,
+        dark_page=COLOR_BG,
+        positive=COLOR_SUCCESS,
+        negative=COLOR_DANGER,
+        info=COLOR_PRIMARY,
+        warning=COLOR_ACCENT,
+    )
+
+
+def _build_header(
+    active: str,
+    brand_subtitle: str,
+    nav_items: list[tuple[str, str]],
+    db_label: str,
+) -> None:
+    """Render the top navigation header.
+
+    Args:
+        active: URL path of the currently active page (e.g. "/").
+        brand_subtitle: Text shown after the brand dot, e.g. "Coevolutivo / GA".
+        nav_items: List of (path, label) tuples for nav links.
+        db_label: Short DB identifier shown in the top-right corner.
+    """
+    with ui.header().classes("items-center justify-between q-px-lg q-py-md"):
+        with ui.row().classes("items-center gap-8"):
+            with ui.row().classes("items-center gap-2").classes("brand"):
+                ui.html('')
+                ui.html(
+                    f'Multi-Swarm / {brand_subtitle}'
+                )
+            with ui.row().classes("items-center gap-1"):
+                for path, label in nav_items:
+                    cls = "nav-link active" if active == path else "nav-link"
+                    ui.link(label, path).classes(cls)
+        with ui.row().classes("items-center gap-3"):
+            ui.html(
+                f'{db_label}'
+            )
diff --git a/src/multi_swarm_core/pyproject.toml b/src/multi_swarm_core/pyproject.toml
index 866c942..18e850d 100644
--- a/src/multi_swarm_core/pyproject.toml
+++ b/src/multi_swarm_core/pyproject.toml
@@ -18,6 +18,8 @@ dependencies = [
     "pyyaml>=6.0",
     "pyarrow>=18.0",
     "yfinance>=1.3.0",
+    "nicegui>=3.11.1",
+    "plotly>=5.24",
 ]
 
 [build-system]
diff --git a/src/strategy_crypto/strategy_crypto/frontend/data.py b/src/strategy_crypto/strategy_crypto/frontend/data.py
index 738a008..0ecb6a7 100644
--- a/src/strategy_crypto/strategy_crypto/frontend/data.py
+++ b/src/strategy_crypto/strategy_crypto/frontend/data.py
@@ -1,3 +1,8 @@
+"""Paper-trading data access functions for the strategy_crypto dashboard.
+
+Reads exclusively from strategy_crypto.db (paper_trading_* tables).
+"""
+
 from __future__ import annotations
 
 import json
@@ -7,53 +12,6 @@ from typing import Any
 
 import pandas as pd  # type: ignore[import-untyped]
 
-from multi_swarm_core.persistence.repository import Repository
-
-
-def get_repo(db_path: str | Path) -> Repository:
-    return Repository(db_path=db_path)
-
-
-def list_runs_df(repo: Repository) -> pd.DataFrame:
-    return pd.DataFrame(repo.list_runs())
-
-
-def get_run_overview(repo: Repository, run_id: str) -> dict[str, Any]:
-    run = repo.get_run(run_id)
-    return {
-        "name": run["name"],
-        "started_at": run["started_at"],
-        "completed_at": run["completed_at"],
-        "status": run["status"],
-        "total_cost_usd": run["total_cost_usd"],
-        "config": json.loads(run["config_json"]),
-    }
-
-
-def generations_df(repo: Repository, run_id: str) -> pd.DataFrame:
-    return pd.DataFrame(repo.list_generations(run_id))
-
-
-def evaluations_df(repo: Repository, run_id: str) -> pd.DataFrame:
-    return pd.DataFrame(repo.list_evaluations(run_id))
-
-
-def genomes_df(
-    repo: Repository, run_id: str, generation_idx: int | None = None
-) -> pd.DataFrame:
-    rows = repo.list_genomes(run_id, generation_idx)
-    flat: list[dict[str, Any]] = []
-    for r in rows:
-        payload = json.loads(r["payload_json"])
-        flat.append(
-            {
-                "id": r["id"],
-                "generation_idx": r["generation_idx"],
-                **payload,
-            }
-        )
-    return pd.DataFrame(flat)
-
 
 def _paper_conn(db_path: str | Path) -> sqlite3.Connection:
     conn = sqlite3.connect(str(db_path))
diff --git a/src/strategy_crypto/strategy_crypto/frontend/nicegui_app.py b/src/strategy_crypto/strategy_crypto/frontend/nicegui_app.py
index eca9201..0f314f2 100644
--- a/src/strategy_crypto/strategy_crypto/frontend/nicegui_app.py
+++ b/src/strategy_crypto/strategy_crypto/frontend/nicegui_app.py
@@ -1,9 +1,7 @@
-"""NiceGUI dashboard — port progressivo da Streamlit.
+"""Strategy Crypto Dashboard — paper-trading page: /.
 
 Avvio: ``uv run python -m strategy_crypto.frontend.nicegui_app``
-Default port 8080. Streamlit resta su 8501 durante la migrazione.
-
-Riusa ``dashboard.data`` (Repository helpers) senza modifiche al backend.
+Default port 8080. Legge SOLO strategy_crypto.db (paper_trading_* tables).
 
 Palette "Neon Trading Dashboard" (ispirata screenshot 2026-05-11):
 - BG: #0A0A0F (near-black con tinge blu)
@@ -18,8 +16,6 @@ Palette "Neon Trading Dashboard" (ispirata screenshot 2026-05-11):
 
 from __future__ import annotations
 
-import html
-import json
 import os
 from pathlib import Path
 from typing import Any
@@ -29,12 +25,6 @@ import plotly.graph_objects as go  # type: ignore[import-untyped]
 from nicegui import app, ui
 
 from strategy_crypto.frontend.data import (
-    evaluations_df,
-    generations_df,
-    genomes_df,
-    get_repo,
-    get_run_overview,
-    list_runs_df,
     paper_equity_df,
     paper_positions_df,
     paper_run_summary,
@@ -42,840 +32,21 @@ from strategy_crypto.frontend.data import (
     paper_ticks_df,
     paper_trades_df,
 )
+from multi_swarm_core.dashboard.theme import (
+    COLOR_PRIMARY,
+    COLOR_SURFACE,
+    COLOR_SURFACE_2,
+    COLOR_TEXT,
+    COLOR_TEXT_MUTED,
+    _STATUS_BADGE,
+    _apply_theme,
+    _build_header,
+)
 
-# Dual-DB: GA core e paper strategy_crypto vivono in DB separati.
-GA_DB_PATH = os.environ.get("GA_DB_PATH", "./state/runs.db")
 PAPER_DB_PATH = os.environ.get("STRATEGY_CRYPTO_DB_PATH", "./state/strategy_crypto.db")
-# Subpath per Traefik: "" in dev, "/strategy_crypto_gui" in prod.
 DASHBOARD_ROOT_PATH = os.environ.get("DASHBOARD_ROOT_PATH", "")
 REFRESH_INTERVAL_S = 3.0
 
-# --- Neon Trading Dashboard palette ---
-COLOR_BG = "#0A0A0F"
-COLOR_SURFACE = "#13131A"
-COLOR_SURFACE_2 = "#1C1C26"
-COLOR_BORDER = "rgba(255, 45, 135, 0.12)"
-COLOR_BORDER_HOVER = "rgba(255, 45, 135, 0.45)"
-COLOR_PRIMARY = "#FF2D87"
-COLOR_SECONDARY = "#00D9FF"
-COLOR_ACCENT = "#FFB800"
-COLOR_SUCCESS = "#00E676"
-COLOR_DANGER = "#FF3D60"
-COLOR_TEXT = "#FFFFFF"
-COLOR_TEXT_MUTED = "#7A7A8C"
-
-_STATUS_BADGE: dict[str, tuple[str, str]] = {
-    "running": ("● running", "positive"),
-    "completed": ("✓ completed", "positive"),
-    "failed": ("✕ failed", "negative"),
-}
-
-_CUSTOM_CSS = f"""
-
-"""
-
-
-def _json_to_html(obj: Any, indent: int = 0) -> str:
-    """Render JSON con span colorati espliciti. Garantisce leggibilità ovunque."""
-    pad = "  " * indent
-    inner_pad = "  " * (indent + 1)
-    if isinstance(obj, dict):
-        if not obj:
-            return '{}'
-        items = []
-        for k, v in obj.items():
-            key = f'"{html.escape(str(k))}"'
-            val = _json_to_html(v, indent + 1)
-            items.append(f"{inner_pad}{key}: {val}")
-        return ('{\n'
-                + ',\n'.join(items)
-                + f'\n{pad}}}')
-    if isinstance(obj, list):
-        if not obj:
-            return '[]'
-        items = [_json_to_html(x, indent + 1) for x in obj]
-        return ('[\n'
-                + ',\n'.join(inner_pad + i for i in items)
-                + f'\n{pad}]')
-    if isinstance(obj, bool):
-        return f'{str(obj).lower()}'
-    if obj is None:
-        return 'null'
-    if isinstance(obj, (int, float)):
-        return f'{obj}'
-    return f'"{html.escape(str(obj))}"'
-
-
-def _apply_theme() -> None:
-    ui.add_head_html(_CUSTOM_CSS)
-    ui.dark_mode().enable()
-    ui.colors(
-        primary=COLOR_PRIMARY,
-        secondary=COLOR_SECONDARY,
-        accent=COLOR_ACCENT,
-        dark=COLOR_BG,
-        dark_page=COLOR_BG,
-        positive=COLOR_SUCCESS,
-        negative=COLOR_DANGER,
-        info=COLOR_PRIMARY,
-        warning=COLOR_ACCENT,
-    )
-
-
-def _build_header(active: str) -> None:
-    with ui.header().classes("items-center justify-between q-px-lg q-py-md"):
-        with ui.row().classes("items-center gap-8"):
-            with ui.row().classes("items-center gap-2").classes("brand"):
-                ui.html('')
-                ui.html('Multi-Swarm / Coevolutivo')
-            with ui.row().classes("items-center gap-1"):
-                for path, label in (
-                    ("/", "Overview"),
-                    ("/convergence", "Convergence"),
-                    ("/genomes", "Genomes"),
-                    ("/paper", "Paper"),
-                ):
-                    cls = "nav-link active" if active == path else "nav-link"
-                    ui.link(label, path).classes(cls)
-        with ui.row().classes("items-center gap-3"):
-            ui.html(f''
-                    f'⛁ {Path(GA_DB_PATH).resolve().name} + {Path(PAPER_DB_PATH).resolve().name}')
-
-
-def _runs_options() -> dict[str, str]:
-    repo = get_repo(GA_DB_PATH)
-    runs = list_runs_df(repo)
-    if runs.empty:
-        return {}
-    return {
-        row["id"]: f"{row['name']} — {row['status']} ({row['started_at'][:16]})"
-        for _, row in runs.iterrows()
-    }
-
-
-def _snapshot(run_id: str) -> dict[str, Any]:
-    repo = get_repo(GA_DB_PATH)
-    ov = get_run_overview(repo, run_id)
-    evals = evaluations_df(repo, run_id)
-    gens = generations_df(repo, run_id)
-
-    cfg = ov["config"]
-    pop_size = int(cfg.get("population_size", 0))
-    n_gens = int(cfg.get("n_generations", 0))
-    evals_total = max(pop_size * n_gens, 1)
-    evals_done = len(evals)
-    gens_done = int(gens["completed_at"].notna().sum()) if not gens.empty else 0
-    # runs.total_cost_usd è 0 finché complete_run non viene chiamato.
-    # Per le run in corso leggiamo la somma live da cost_records.
-    live_cost = float(repo.total_cost(run_id)) if ov["status"] == "running" else float(
-        ov["total_cost_usd"]
-    )
-
-    top_fit = float(evals["fitness"].max()) if evals_done else float("nan")
-    median_fit = float(evals["fitness"].median()) if evals_done else float("nan")
-    parse_success = (
-        100.0 * float(evals["parse_error"].isna().sum()) / evals_done if evals_done else 0.0
-    )
-
-    return {
-        "status": ov["status"],
-        "name": cfg.get("run_name", "—"),
-        "started_at": ov["started_at"],
-        "completed_at": ov["completed_at"] or "—",
-        "cost_usd": live_cost,
-        "pop_size": pop_size,
-        "n_gens": n_gens,
-        "evals_done": evals_done,
-        "evals_total": evals_total,
-        "gens_done": gens_done,
-        "top_fit": top_fit,
-        "median_fit": median_fit,
-        "parse_success": parse_success,
-        "config": cfg,
-        "gens_df": gens,
-    }
-
-
-def _convergence_figure(gens_df: Any) -> go.Figure:
-    fig = go.Figure()
-    if gens_df.empty:
-        fig.add_annotation(
-            text="Nessuna generazione registrata", x=0.5, y=0.5, showarrow=False,
-            font={"color": COLOR_TEXT_MUTED, "size": 14},
-        )
-    else:
-        fig.add_trace(
-            go.Scatter(
-                x=gens_df["generation_idx"], y=gens_df["fitness_max"],
-                name="max", mode="lines+markers",
-                line={"color": COLOR_PRIMARY, "width": 3, "shape": "spline", "smoothing": 0.6},
-                marker={"size": 9, "color": COLOR_PRIMARY,
-                        "line": {"color": "#fff", "width": 1}},
-                fill="tozeroy",
-                fillcolor="rgba(255, 45, 135, 0.12)",
-            )
-        )
-        fig.add_trace(
-            go.Scatter(
-                x=gens_df["generation_idx"], y=gens_df["fitness_p90"],
-                name="p90", mode="lines+markers",
-                line={"color": COLOR_ACCENT, "width": 2, "dash": "dot", "shape": "spline"},
-                marker={"size": 7, "color": COLOR_ACCENT},
-            )
-        )
-        fig.add_trace(
-            go.Scatter(
-                x=gens_df["generation_idx"], y=gens_df["fitness_median"],
-                name="median", mode="lines+markers",
-                line={"color": COLOR_SECONDARY, "width": 2, "shape": "spline"},
-                marker={"size": 7, "color": COLOR_SECONDARY},
-            )
-        )
-    fig.update_layout(
-        template="plotly_dark",
-        paper_bgcolor=COLOR_SURFACE,
-        plot_bgcolor=COLOR_SURFACE,
-        font={"color": COLOR_TEXT},
-        xaxis={"title": "generation", "gridcolor": "rgba(148, 163, 184, 0.08)", "dtick": 1},
-        yaxis={"title": "fitness", "gridcolor": "rgba(148, 163, 184, 0.08)"},
-        title={"text": "Fitness convergence", "font": {"color": COLOR_TEXT, "size": 18}},
-        legend={"bgcolor": "rgba(19, 19, 26, 0.95)", "bordercolor": COLOR_PRIMARY, "borderwidth": 1},
-        margin={"l": 50, "r": 30, "t": 50, "b": 50},
-    )
-    return fig
-
-
-def _entropy_figure(gens_df: Any) -> go.Figure:
-    fig = go.Figure()
-    if not gens_df.empty:
-        fig.add_trace(
-            go.Scatter(
-                x=gens_df["generation_idx"], y=gens_df["entropy"],
-                mode="lines+markers",
-                line={"color": COLOR_SECONDARY, "width": 3, "shape": "spline", "smoothing": 0.6},
-                marker={"size": 9, "color": COLOR_SECONDARY,
-                        "line": {"color": "#fff", "width": 1}},
-                fill="tozeroy",
-                fillcolor="rgba(0, 217, 255, 0.12)",
-                name="entropy",
-            )
-        )
-        fig.add_hline(
-            y=0.5, line_dash="dash", line_color=COLOR_ACCENT,
-            annotation_text="gate threshold (0.5)",
-            annotation_font_color=COLOR_ACCENT,
-        )
-    fig.update_layout(
-        template="plotly_dark",
-        paper_bgcolor=COLOR_SURFACE,
-        plot_bgcolor=COLOR_SURFACE,
-        font={"color": COLOR_TEXT},
-        xaxis={"title": "generation", "gridcolor": "rgba(148, 163, 184, 0.08)", "dtick": 1},
-        yaxis={"title": "entropy", "gridcolor": "rgba(148, 163, 184, 0.08)"},
-        title={"text": "Diversity (fitness entropy)", "font": {"color": COLOR_TEXT, "size": 18}},
-        margin={"l": 50, "r": 30, "t": 50, "b": 50},
-    )
-    return fig
-
-
-@ui.page("/")
-def index() -> None:
-    _apply_theme()
-    _build_header(active="/")
-
-    options = _runs_options()
-    if not options:
-        ui.label("Nessuna run nel database.").classes("text-h5")
-        return
-
-    state: dict[str, Any] = {"run_id": next(iter(options))}
-
-    with ui.row().classes("w-full items-center gap-4 q-mb-md"):
-        select = ui.select(options=options, value=state["run_id"], label="Run").classes(
-            "flex-grow"
-        )
-        status_badge = ui.badge("…", color="primary").classes("text-body1 q-pa-sm")
-        ui.button("🔄 Refresh", on_click=lambda: refresh()).props("outline color=primary")
-
-    with ui.card().classes("w-full"):
-        ui.label("Progresso run").classes("text-subtitle1")
-        gen_label = ui.label("Generations: 0/0")
-        gen_bar = ui.linear_progress(0.0, show_value=False).props("size=20px color=primary")
-        eval_label = ui.label("Evaluations: 0/0 (0.0%)")
-        eval_bar = ui.linear_progress(0.0, show_value=False).props("size=20px color=accent")
-
-    with ui.row().classes("w-full gap-4"):
-        with ui.card().classes("flex-grow metric-card accent-cyan"):
-            ui.label("Top fitness").classes("text-caption")
-            top_lbl = ui.label("—").classes("text-h4")
-        with ui.card().classes("flex-grow metric-card accent-purple"):
-            ui.label("Median fitness").classes("text-caption")
-            median_lbl = ui.label("—").classes("text-h4")
-        with ui.card().classes("flex-grow metric-card accent-amber"):
-            ui.label("Parse success").classes("text-caption")
-            parse_lbl = ui.label("—").classes("text-h4")
-        with ui.card().classes("flex-grow metric-card accent-green"):
-            ui.label("Cost (USD)").classes("text-caption")
-            cost_lbl = ui.label("—").classes("text-h4")
-
-    with ui.row().classes("w-full gap-4 q-mt-md"):
-        started_lbl = ui.label("Started: —")
-        completed_lbl = ui.label("Completed: —")
-    ui.separator()
-    ui.label("Config").classes("text-subtitle1")
-    cfg_code = ui.html('
').classes("w-full")
-
-    def refresh() -> None:
-        run_id = select.value
-        if not run_id:
-            return
-        try:
-            s = _snapshot(run_id)
-        except Exception as e:  # noqa: BLE001
-            ui.notify(f"Errore: {e}", type="negative")
-            return
-
-        text, color = _STATUS_BADGE.get(s["status"], (s["status"], "primary"))
-        status_badge.text = text
-        status_badge.props(f"color={color}")
-
-        gen_frac = min(s["gens_done"] / max(s["n_gens"], 1), 1.0)
-        eval_frac = min(s["evals_done"] / s["evals_total"], 1.0)
-        gen_bar.value = gen_frac
-        eval_bar.value = eval_frac
-        gen_label.text = f"Generations: {s['gens_done']}/{s['n_gens']}"
-        eval_label.text = (
-            f"Evaluations: {s['evals_done']}/{s['evals_total']} ({100 * eval_frac:.1f}%)"
-        )
-
-        top_lbl.text = f"{s['top_fit']:.4f}" if s["evals_done"] else "—"
-        median_lbl.text = f"{s['median_fit']:.4f}" if s["evals_done"] else "—"
-        parse_lbl.text = f"{s['parse_success']:.1f}%" if s["evals_done"] else "—"
-        cost_lbl.text = f"${s['cost_usd']:.4f}"
-
-        started_lbl.text = f"Started: {s['started_at']}"
-        completed_lbl.text = f"Completed: {s['completed_at']}"
-        cfg_code.content = f'
{_json_to_html(s["config"])}
'
-
-    def on_select_change() -> None:
-        state["run_id"] = select.value
-        refresh()
-
-    select.on_value_change(on_select_change)
-    ui.timer(REFRESH_INTERVAL_S, refresh)
-    refresh()
-
-
-@ui.page("/convergence")
-def convergence() -> None:
-    _apply_theme()
-    _build_header(active="/convergence")
-
-    options = _runs_options()
-    if not options:
-        ui.label("Nessuna run nel database.").classes("text-h5")
-        return
-
-    state: dict[str, Any] = {"run_id": next(iter(options))}
-
-    with ui.row().classes("w-full items-center gap-4 q-mb-md"):
-        select = ui.select(options=options, value=state["run_id"], label="Run").classes(
-            "flex-grow"
-        )
-        gen_count_lbl = ui.label("Gens: 0/0").classes("text-body1").style(
-            f"color: {COLOR_PRIMARY}; font-weight: 600;"
-        )
-        ui.button("🔄 Refresh", on_click=lambda: refresh()).props("outline color=primary")
-
-    fitness_plot = ui.plotly(_convergence_figure(generations_df(get_repo(GA_DB_PATH), state["run_id"]))).classes("w-full")
-    entropy_plot = ui.plotly(_entropy_figure(generations_df(get_repo(GA_DB_PATH), state["run_id"]))).classes("w-full q-mt-md")
-
-    ui.separator()
-    ui.label("Tabella generazioni").classes("text-subtitle1 q-mt-md")
-    gens_table = ui.table(
-        columns=[
-            {"name": "generation_idx", "label": "gen", "field": "generation_idx", "sortable": True},
-            {"name": "n_genomes", "label": "n", "field": "n_genomes"},
-            {"name": "fitness_max", "label": "max", "field": "fitness_max"},
-            {"name": "fitness_p90", "label": "p90", "field": "fitness_p90"},
-            {"name": "fitness_median", "label": "median", "field": "fitness_median"},
-            {"name": "entropy", "label": "entropy", "field": "entropy"},
-            {"name": "completed_at", "label": "completed", "field": "completed_at"},
-        ],
-        rows=[],
-        row_key="generation_idx",
-    ).classes("w-full")
-
-    def refresh() -> None:
-        run_id = select.value
-        if not run_id:
-            return
-        try:
-            gens = generations_df(get_repo(GA_DB_PATH), run_id)
-            ov = get_run_overview(get_repo(GA_DB_PATH), run_id)
-        except Exception as e:  # noqa: BLE001
-            ui.notify(f"Errore: {e}", type="negative")
-            return
-
-        n_gens = int(ov["config"].get("n_generations", 0))
-        gens_done = int(gens["completed_at"].notna().sum()) if not gens.empty else 0
-        gen_count_lbl.text = f"Gens: {gens_done}/{n_gens}"
-
-        fitness_plot.update_figure(_convergence_figure(gens))
-        entropy_plot.update_figure(_entropy_figure(gens))
-
-        if gens.empty:
-            gens_table.rows = []
-        else:
-            display_cols = [
-                "generation_idx", "n_genomes",
-                "fitness_max", "fitness_p90", "fitness_median",
-                "entropy", "completed_at",
-            ]
-            gens_table.rows = [
-                {
-                    col: (round(v, 6) if isinstance(v, float) else v)
-                    for col, v in row.items()
-                    if col in display_cols
-                }
-                for _, row in gens.iterrows()
-            ]
-        gens_table.update()
-
-    def on_select_change() -> None:
-        state["run_id"] = select.value
-        refresh()
-
-    select.on_value_change(on_select_change)
-    ui.timer(REFRESH_INTERVAL_S, refresh)
-    refresh()
-
-
-@ui.page("/genomes")
-def genomes() -> None:
-    _apply_theme()
-    _build_header(active="/genomes")
-
-    options = _runs_options()
-    if not options:
-        ui.label("Nessuna run nel database.").classes("text-h5")
-        return
-
-    state: dict[str, Any] = {
-        "run_id": next(iter(options)),
-        "selected_gid": None,
-        "merged": None,
-    }
-
-    with ui.row().classes("w-full items-center gap-4 q-mb-md"):
-        select = ui.select(options=options, value=state["run_id"], label="Run").classes(
-            "flex-grow"
-        )
-        top_k_select = ui.select(
-            options={10: "Top 10", 25: "Top 25", 50: "Top 50"},
-            value=10,
-            label="Top K",
-        )
-        ui.button("🔄 Refresh", on_click=lambda: refresh()).props("outline color=primary")
-
-    ui.label("Top genomi per fitness").classes("text-subtitle1 q-mt-sm")
-    top_table = ui.table(
-        columns=[
-            {"name": "genome_id", "label": "id", "field": "genome_id", "align": "left"},
-            {"name": "fitness", "label": "fitness", "field": "fitness", "sortable": True},
-            {"name": "dsr", "label": "DSR", "field": "dsr"},
-            {"name": "sharpe", "label": "Sharpe", "field": "sharpe"},
-            {"name": "max_dd", "label": "max DD", "field": "max_dd"},
-            {"name": "n_trades", "label": "trades", "field": "n_trades"},
-            {"name": "cognitive_style", "label": "style", "field": "cognitive_style"},
-            {"name": "temperature", "label": "T", "field": "temperature"},
-            {"name": "lookback_window", "label": "lookback", "field": "lookback_window"},
-        ],
-        rows=[],
-        row_key="genome_id",
-        selection="single",
-    ).classes("w-full")
-
-    ui.separator().classes("q-my-md")
-
-    with ui.card().classes("w-full"):
-        ui.label("Ispezione genoma").classes("text-subtitle1")
-        detail_hint = ui.label("Seleziona un genoma dalla tabella sopra.").classes(
-            "text-caption"
-        ).style(f"color: {COLOR_TEXT_MUTED};")
-
-        with ui.row().classes("w-full gap-4 q-mt-sm"):
-            with ui.card().classes("flex-grow metric-card accent-cyan"):
-                ui.label("fitness").classes("text-caption")
-                fit_lbl = ui.label("—").classes("text-h4")
-            with ui.card().classes("flex-grow metric-card accent-purple"):
-                ui.label("DSR").classes("text-caption")
-                dsr_lbl = ui.label("—").classes("text-h4")
-            with ui.card().classes("flex-grow metric-card accent-amber"):
-                ui.label("Sharpe").classes("text-caption")
-                sharpe_lbl = ui.label("—").classes("text-h4")
-            with ui.card().classes("flex-grow metric-card"):
-                ui.label("max DD").classes("text-caption")
-                dd_lbl = ui.label("—").classes("text-h4")
-            with ui.card().classes("flex-grow metric-card accent-green"):
-                ui.label("trades").classes("text-caption")
-                trades_lbl = ui.label("—").classes("text-h4")
-            with ui.card().classes("flex-grow metric-card"):
-                ui.label("style").classes("text-caption")
-                style_lbl = ui.label("—").classes("text-h4")
-
-        ui.label("System prompt").classes("text-subtitle1 q-mt-md")
-        prompt_code = ui.html('
—
').classes("w-full")
-
-        ui.label("Raw LLM output").classes("text-subtitle1 q-mt-md")
-        raw_code = ui.html('
—
').classes("w-full")
-
-        parse_error_lbl = ui.label("").classes("q-mt-sm").style(
-            "color: #FF6B6B; font-weight: 600;"
-        )
-
-    def _render_detail(row: dict[str, Any]) -> None:
-        detail_hint.text = f"Genoma: {row.get('genome_id', '—')}"
-        fit_lbl.text = f"{float(row.get('fitness', 0)):.4f}"
-        dsr_lbl.text = f"{float(row.get('dsr', 0)):.4f}"
-        sharpe_lbl.text = f"{float(row.get('sharpe', 0)):.3f}"
-        dd_lbl.text = f"{float(row.get('max_dd', 0)):.3f}"
-        trades_lbl.text = str(int(row.get("n_trades", 0)))
-        style_lbl.text = str(row.get("cognitive_style", "—"))
-        prompt_code.content = (
-            f'
{html.escape(str(row.get("system_prompt", "—")))}
'
-        )
-        raw_code.content = (
-            f'
{html.escape(str(row.get("raw_text", "—") or "—"))}
'
-        )
-        pe = row.get("parse_error")
-        parse_error_lbl.text = f"❌ Parse error: {pe}" if pe else ""
-
-    def refresh() -> None:
-        run_id = select.value
-        if not run_id:
-            return
-        try:
-            repo = get_repo(GA_DB_PATH)
-            evals = evaluations_df(repo, run_id)
-            gens = genomes_df(repo, run_id)
-        except Exception as e:  # noqa: BLE001
-            ui.notify(f"Errore: {e}", type="negative")
-            return
-
-        if evals.empty:
-            top_table.rows = []
-            top_table.update()
-            return
-
-        merged = evals.merge(
-            gens, left_on="genome_id", right_on="id", how="left", suffixes=("", "_g")
-        )
-        state["merged"] = merged
-
-        k = int(top_k_select.value)
-        top = merged.sort_values("fitness", ascending=False).head(k)
-
-        rows = []
-        for _, r in top.iterrows():
-            rows.append(
-                {
-                    "genome_id": str(r.get("genome_id", "—"))[:12] + "…",
-                    "fitness": round(float(r.get("fitness", 0)), 4),
-                    "dsr": round(float(r.get("dsr", 0)), 4),
-                    "sharpe": round(float(r.get("sharpe", 0)), 3),
-                    "max_dd": round(float(r.get("max_dd", 0)), 3),
-                    "n_trades": int(r.get("n_trades", 0)),
-                    "cognitive_style": str(r.get("cognitive_style", "—")),
-                    "temperature": round(float(r.get("temperature", 0)), 2),
-                    "lookback_window": int(r.get("lookback_window", 0)),
-                    "_full_id": str(r.get("genome_id", "")),
-                }
-            )
-        top_table.rows = rows
-        top_table.update()
-
-        sel = state.get("selected_gid")
-        if sel:
-            match = merged[merged["genome_id"] == sel]
-            if not match.empty:
-                _render_detail(match.iloc[0].to_dict())
-
-    def on_row_selected(e: Any) -> None:
-        rows = (e.args or {}).get("rows") or []
-        if not rows:
-            return
-        full_id = rows[0].get("_full_id")
-        if not full_id:
-            return
-        state["selected_gid"] = full_id
-        merged = state.get("merged")
-        if merged is None:
-            return
-        match = merged[merged["genome_id"] == full_id]
-        if not match.empty:
-            _render_detail(match.iloc[0].to_dict())
-
-    def on_select_change() -> None:
-        state["run_id"] = select.value
-        state["selected_gid"] = None
-        refresh()
-
-    select.on_value_change(on_select_change)
-    top_k_select.on_value_change(lambda _: refresh())
-    top_table.on("selection", on_row_selected)
-    ui.timer(REFRESH_INTERVAL_S, refresh)
-    refresh()
-
 
 def _paper_runs_options(only_running: bool = False) -> dict[str, str]:
     runs = paper_runs_df(PAPER_DB_PATH)
@@ -925,10 +96,15 @@ def _paper_equity_figure(eq_df: Any, initial_capital: float) -> go.Figure:
     return fig
 
 
-@ui.page("/paper")
+@ui.page("/")
 def paper() -> None:
     _apply_theme()
-    _build_header(active="/paper")
+    _build_header(
+        active="/",
+        brand_subtitle="Strategy Crypto",
+        nav_items=[("/", "Paper")],
+        db_label=f"⛁ {Path(PAPER_DB_PATH).resolve().name}",
+    )
 
     options = _paper_runs_options()
     if not options:
@@ -1087,7 +263,6 @@ def paper() -> None:
 def main() -> None:
     app.on_startup(
         lambda: print(
-            f"GA DB: {Path(GA_DB_PATH).resolve()} | "
             f"Paper DB: {Path(PAPER_DB_PATH).resolve()} | "
             f"root_path: {DASHBOARD_ROOT_PATH or '/'}"
         )
diff --git a/uv.lock b/uv.lock
index c64a9b6..d2e48fa 100644
--- a/uv.lock
+++ b/uv.lock
@@ -936,9 +936,11 @@ version = "0.1.0"
 source = { editable = "src/multi_swarm_core" }
 dependencies = [
     { name = "httpx" },
+    { name = "nicegui" },
     { name = "numpy" },
     { name = "openai" },
     { name = "pandas" },
+    { name = "plotly" },
     { name = "pyarrow" },
     { name = "pydantic" },
     { name = "pydantic-settings" },
@@ -953,9 +955,11 @@ dependencies = [
 [package.metadata]
 requires-dist = [
     { name = "httpx", specifier = ">=0.28" },
+    { name = "nicegui", specifier = ">=3.11.1" },
     { name = "numpy", specifier = ">=2.1" },
     { name = "openai", specifier = ">=1.55" },
     { name = "pandas", specifier = ">=2.2" },
+    { name = "plotly", specifier = ">=5.24" },
     { name = "pyarrow", specifier = ">=18.0" },
     { name = "pydantic", specifier = ">=2.9" },
     { name = "pydantic-settings", specifier = ">=2.6" },