merge: dev2 → master — reset full-date (epoch fonte di verità, web+cloud) + bridge tick 3-sezioni + diario Capitano + Sentinella stabilizzazione

# Conflicts:
#	.launcher/sentinel-bridge.py

Author: leopu00

.launcher/pacing-bridge.py

@@ -891,7 +891,7 @@ def format_message(d: dict) -> str:
             extra = f" delta_usage={d['delta_usage']} team_kt={d.get('team_kt', '?')}"
         return f"[BRIDGE PACING] tick saltato reason={why}{extra}."
 
-    ts = d["now"].strftime("%H:%M UTC")
+    ts = d["now"].strftime("%Y-%m-%d %H:%M UTC")
     usage_now = d["usage_now"]
     proj = d["proj"] if d["proj"] is not None else "?"
     reset_at = d["reset_at"] or "?"
@@ -906,7 +906,7 @@ def format_message(d: dict) -> str:
     )
     parts = [
         f"[BRIDGE PACING] {ts} window={eff_str} samples={d['n_samples']}",
-        f"usage={usage_now}% reset_in={h_str} reset_at={reset_at}UTC "
+        f"usage={usage_now}% reset_in={h_str} reset_at={reset_at} "
         f"(proj={proj}% — INFO, segnale secondario volatile: NON guida le "
         f"decisioni, usa vel_team vs vel_target)",
         f"vel_team={d['vel_team']:.2f}%/h",

.launcher/sentinel-bridge.py

@@ -0,0 +1,53 @@
+---
+name: captain-diary
+description: Daily handoff diary for the Captain. The Captain is restarted often (context-refresh, new work window, reboot) and otherwise loses the day's hard-won pacing lessons — repeating the same mistakes (e.g. 3 Scouts at once → an unbrakable spike → a 5h coast to repay the debt). Read the PREVIOUS day's notes at startup (handoff), and APPEND a one-line note whenever something significant happens during the day (a scaling decision, a spike, a kill, a lesson). One append-only file per day.
+allowed-tools: Bash(python3 /app/shared/skills/captain_diary.py *)
+---
+
+# captain-diary — passaggio del testimone tra Capitani
+
+Un file per giorno in `$JHT_HOME/logs/captain-diary-YYYY-MM-DD.md`, append-only.
+Serve a **non ripartire da zero ogni riavvio**: le lezioni di pacing di oggi
+passano al Capitano di domani.
+
+## Al risveglio (SEMPRE, prima di lavorare)
+
+Leggi le note del Capitano del giorno precedente:
+
+```bash
+python3 /app/shared/skills/captain_diary.py handoff
+```
+
+Stampa le note di **ieri** (o dell'ultimo giorno lavorato) + ciò che è già
+annotato **oggi**. Erediti le lezioni → **non rifare gli stessi errori**. Se non
+c'è nulla, sei il primo: inizia ad annotare.
+
+## Durante il giorno — annota gli eventi SIGNIFICATIVI
+
+Una riga, quando succede qualcosa da cui si impara. NON il diario di tutto:
+solo ciò che servirebbe al Capitano di domani.
+
+```bash
+python3 /app/shared/skills/captain_diary.py add "20:05 — 3 Scout insieme: picco \
+infrenabile in 15 min, 5h di coast per ripagare il debito. Lezione: max 1 Scout \
+poi 30 min di osservazione (C-02)."
+```
+
+Cosa vale la pena annotare:
+- decisioni di scaling che sono andate male (o bene) — quanti worker, che throttle, cosa è successo;
+- un picco non frenabile e come l'hai recuperato;
+- un kill e perché;
+- un pattern emerso (es. "lo Scout sul sito X consuma il doppio");
+- qualunque cosa che, se la sapessi domani, eviterebbe un errore.
+
+## Rivedere solo oggi
+
+```bash
+python3 /app/shared/skills/captain_diary.py today
+```
+
+## Regola
+
+- Il diario è il **testimone**: leggilo al boot, alimentalo durante il giorno.
+- Note **brevi e azionabili** (un fatto + la lezione), non un log verboso.
+- L'orario lo mette lo strumento: tu scrivi solo il fatto e la lezione.

agents/_skills/captain-diary/SKILL.md

@@ -21,7 +21,7 @@ Typische Ausgabe:
 ```
 === Rate Budget — claude ===
   Usage:            53%
-  Reset:            13:49 (in 2h 34m)
+  Reset:            tra 2h 34m (2026-04-24 15:49 CEST)
   Measured velocity:+0.39%/h (EMA)
   Target velocity:  11.38%/h (to close at 92% by reset)
   Reset projection: 56%
@@ -46,7 +46,7 @@ Typische Ausgabe:
 
 ```bash
 python3 /app/shared/skills/rate_budget.py status
-# → provider=claude usage=55% status=OK throttle=T0 reset=13:49 (in 2h 34m)
+# → provider=claude usage=55% status=OK throttle=0 reset_in=2h 34m (at 2026-04-24 15:49 CEST)
 ```
 
 Nützlich für schnelle Logs oder Prüfungen mitten in der Schleife.

agents/_skills/rate-budget/SKILL.de.md

@@ -21,7 +21,7 @@ Salida típica:
 ```
 === Rate Budget — claude ===
   Usage:            53%
-  Reset:            13:49 (in 2h 34m)
+  Reset:            tra 2h 34m (2026-04-24 15:49 CEST)
   Measured velocity:+0.39%/h (EMA)
   Target velocity:  11.38%/h (to close at 92% by reset)
   Reset projection: 56%
@@ -46,7 +46,7 @@ Salida típica:
 
 ```bash
 python3 /app/shared/skills/rate_budget.py status
-# → provider=claude usage=55% status=OK throttle=T0 reset=13:49 (in 2h 34m)
+# → provider=claude usage=55% status=OK throttle=0 reset_in=2h 34m (at 2026-04-24 15:49 CEST)
 ```
 
 Útil para logs rápidos o comprobaciones a mitad de bucle.

agents/_skills/rate-budget/SKILL.es.md

@@ -21,7 +21,7 @@ Sortie typique :
 ```
 === Rate Budget — claude ===
   Usage:            53%
-  Reset:            13:49 (in 2h 34m)
+  Reset:            tra 2h 34m (2026-04-24 15:49 CEST)
   Measured velocity:+0.39%/h (EMA)
   Target velocity:  11.38%/h (to close at 92% by reset)
   Reset projection: 56%
@@ -46,7 +46,7 @@ Sortie typique :
 
 ```bash
 python3 /app/shared/skills/rate_budget.py status
-# → provider=claude usage=55% status=OK throttle=T0 reset=13:49 (in 2h 34m)
+# → provider=claude usage=55% status=OK throttle=0 reset_in=2h 34m (at 2026-04-24 15:49 CEST)
 ```
 
 Utile pour des logs rapides ou des vérifications en cours de boucle.

agents/_skills/rate-budget/SKILL.fr.md

@@ -21,7 +21,7 @@ Tipikus kimenet:
 ```
 === Rate Budget — claude ===
   Usage:            53%
-  Reset:            13:49 (in 2h 34m)
+  Reset:            tra 2h 34m (2026-04-24 15:49 CEST)
   Measured velocity:+0.39%/h (EMA)
   Target velocity:  11.38%/h (to close at 92% by reset)
   Reset projection: 56%
@@ -46,7 +46,7 @@ Tipikus kimenet:
 
 ```bash
 python3 /app/shared/skills/rate_budget.py status
-# → provider=claude usage=55% status=OK throttle=T0 reset=13:49 (in 2h 34m)
+# → provider=claude usage=55% status=OK throttle=0 reset_in=2h 34m (at 2026-04-24 15:49 CEST)
 ```
 
 Hasznos gyors logokhoz vagy ciklus közbeni ellenőrzésekhez.

agents/_skills/rate-budget/SKILL.hu.md

@@ -21,7 +21,7 @@ Output tipico:
 ```
 === Rate Budget — claude ===
   Usage:            53%
-  Reset:            13:49 (in 2h 34m)
+  Reset:            tra 2h 34m (2026-04-24 15:49 CEST)
   Measured velocity:+0.39%/h (EMA)
   Target velocity:  11.38%/h (to close at 92% by reset)
   Reset projection: 56%
@@ -46,7 +46,7 @@ Output tipico:
 
 ```bash
 python3 /app/shared/skills/rate_budget.py status
-# → provider=claude usage=55% status=OK throttle=T0 reset=13:49 (in 2h 34m)
+# → provider=claude usage=55% status=OK throttle=0 reset_in=2h 34m (at 2026-04-24 15:49 CEST)
 ```
 
 Utile per log veloci o check a metà loop.

agents/_skills/rate-budget/SKILL.it.md

@@ -8,47 +8,59 @@ allowed-tools: Bash(python3 *)
 
 The monitoring bridge (`.launcher/sentinel-bridge.py`) polls the active provider every 1–10 min (dynamic — more often under pressure) and writes each sample to `/jht_home/logs/sentinel-data.jsonl`. This skill reads only the **latest sample** that's already been written — no extra provider call.
 
-## At Captain startup
+## Primary: the SAME tick the Sentinella gets
 
-Before spawning any agent, run:
+Run with no argument (or `tick`):
 
 ```bash
-python3 /app/shared/skills/rate_budget.py plan
+python3 /app/shared/skills/rate_budget.py tick
 ```
 
-Typical output:
+You get the **exact same message** the bridge sends to the Sentinella — the
+bridge renders it every tick into `logs/last-tick.txt` and this reads it back
+(zero provider calls, zero divergence). Three airy sections + a bridge hint:
+
 ```
-=== Rate Budget — claude ===
-  Usage:            53%
-  Reset:            13:49 (in 2h 34m)
-  Measured velocity:+0.39%/h (EMA)
-  Target velocity:  11.38%/h (to close at 92% by reset)
-  Reset projection: 56%
-  Status:           OK
-  Throttle:         T0 full speed
-  Host:             cpu=4.7% ram=9.8% (OK)
-
-  Recommended policy: Spawn freely in parallel — keep normal pace.
-  Margin to 92% target: 39%
-  Last tick:        2026-04-24T10:23:18.705062+00:00
+── BRIDGE TICK · 12:45:03 · codex · ON ──
+
+⏱  FINESTRA 5H
+   usage 51%        target 92%  →  chiusura 2026-06-30 16:45 CEST (tra 4h 00m)
+   velocità  11.5 %/h   ·   target  10.2 %/h     [+12% sopra]
+   → ATTENZIONE   ·   proj 97%
+
+📅  OGGI  (budget giornaliero)
+   consumato 4.2% / budget 9.0% (cap 14.0%)   ✅
+   velocità  1.1 %/h   ·   target  0.8 %/h     [+38% sopra]
+
+📆  SETTIMANA
+   usato 16%   ·   rimane 84%   ·   reset 2026-07-07 05:00 CEST (tra 6g 16h)
+   velocità  3.4 %/h   ·   sostenibile  1.1 %/h     [+209% sopra]   ratio 3.09×
+   SOPRA-PACE   ·   debito +6pp   ·   lockout ~58h
+
+🧭  CONSIGLIO BRIDGE
+   Sopra-pace: throttle-to-pace + STOP nuovi spawn finché rientri. ...
+src=bridge.
 ```
 
-**Captain interpretation** (usa `Measured velocity` vs `Target velocity` — NON `Reset projection`, che è INFO volatile):
-- `Throttle T0–T1` + `Measured velocity` ben sotto `Target velocity` (under-pace) → full spawn (Scout + Analyst + Scorer + Writer + Critic)
-- `Throttle T1–T2` + `Measured` ≈ `Target` (on-pace) → reduced spawn (one instance per role)
-- `Throttle T2+` or `Measured velocity` sopra `Target velocity` (burning) → **no spawn**, wait for the bridge to clear the throttle
-- `Reset projection` è solo INFO (estrapolazione volatile a fine finestra) — non basare lo spawn su quello.
+**Read it like this — la velocità è IL segnale** (proj è solo INFO volatile):
+- ogni sezione ha **velocità attuale** (usage consumato / tempo trascorso) vs **target** (quanto manca / tempo che resta), stessa unità %/h. Lo scarto `[+X%/-X%]` è il giudizio.
+- **5h** = il soffitto del provider (lock a 100% prima del reset). **Oggi** = il budget del giorno. **Settimana** = budget + debito.
+- velocità **sotto** target su tutte → c'è margine → spawn graduale (C-02). Velocità **sopra** target → frena / non spawnare. Segui il `🧭 CONSIGLIO`, ma **decidi tu** (C-01).
 
-**If output is `NO_DATA`:** the bridge hasn't polled yet. Wait 1–2 min and retry. Do not start the team without this signal — you risk saturating the rate-limit blind.
+**If output is `NO_DATA`:** the bridge hasn't written a tick yet (not started, or off-hours). Wait 1–2 min and retry.
 
-## One-liner version (scriptable)
+## Other modes
 
 ```bash
-python3 /app/shared/skills/rate_budget.py status
-# → provider=claude usage=55% status=OK throttle=T0 reset=13:49 (in 2h 34m)
+python3 /app/shared/skills/rate_budget.py plan     # scheda dettagliata (velocity EMA, host, policy)
+python3 /app/shared/skills/rate_budget.py status   # one-liner scriptable
+python3 /app/shared/skills/rate_budget.py live      # FORZA un fetch fresco al provider (costa) — solo per dubbi gravi
 ```
 
-Useful for quick logs or mid-loop checks.
+`status` one-liner:
+```
+# → provider=claude usage=55% status=OK throttle=0 reset_in=2h 34m (at 2026-04-24 15:49 CEST)
+```
 
 ## When NOT to use it
 

agents/_skills/rate-budget/SKILL.md

@@ -21,7 +21,7 @@ Saída típica:
 ```
 === Rate Budget — claude ===
   Usage:            53%
-  Reset:            13:49 (in 2h 34m)
+  Reset:            tra 2h 34m (2026-04-24 15:49 CEST)
   Measured velocity:+0.39%/h (EMA)
   Target velocity:  11.38%/h (to close at 92% by reset)
   Reset projection: 56%
@@ -46,7 +46,7 @@ Saída típica:
 
 ```bash
 python3 /app/shared/skills/rate_budget.py status
-# → provider=claude usage=55% status=OK throttle=T0 reset=13:49 (in 2h 34m)
+# → provider=claude usage=55% status=OK throttle=0 reset_in=2h 34m (at 2026-04-24 15:49 CEST)
 ```
 
 Útil para logs rápidos ou verificações a meio do ciclo.