| name | check-usage-tui |
| description | Check provider usage via TUI worker (~30s, costs CLI tokens). FALLBACK when `check-usage-http` returns RATE_LIMIT or fails. Spawns an ephemeral tmux session, runs `/usage` (or `/status` for Codex), parses the modal, kills the session. |
| allowed-tools | Bash(tmux *), Bash(python3 *) |
Skill โ Check usage via TUI worker (robusto, ~30s, costoso in token)
QUANDO USARLA
Solo come L2 dopo che L1 (HTTP) ha fallito. ร robusta perchรฉ bypassa il rate-limit Anthropic (legge dal CLI locale), ma costa:
- ~30s di tempo (spawn worker + boot CLI + render modal)
- alcuni token Claude (apertura sessione + /usage)
- rischio di TUI in stato anomalo (Loading, scaduto, etc.)
Ordine di fallback: HTTP โ TUI โ FATAL.
Workflow
Step 1 โ Provider attivo
python3 -c "import json; print(json.load(open('/jht_home/jht.config.json'))['active_provider'])"
Step 2 โ Spawn sessione tmux EFFIMERA
Nome unico col timestamp (no collisione):
SESS="SENTINELLA-WORKER-$(date +%s)"
tmux new-session -d -x 220 -y 50 -s "$SESS" -c /jht_home
tmux send-keys -t "$SESS" "export HOME='/jht_home'" C-m
tmux send-keys -t "$SESS" "export PATH='/app/agents/_tools:/jht_home/.npm-global/bin:\$PATH'" C-m
tmux send-keys -t "$SESS" "export KIMI_CLI_NO_AUTO_UPDATE=1" C-m
Step 3 โ Lancia il CLI provider
| Provider | Comando |
|---|
claude / anthropic | claude --dangerously-skip-permissions |
kimi / moonshot | kimi --yolo |
codex / openai | codex --yolo |
tmux send-keys -t "$SESS" "<comando>" C-m
sleep 18
Step 4 โ Slash command
| Provider | Slash command |
|---|
claude / anthropic | /usage |
kimi / moonshot | /usage |
codex / openai | /status (NON /usage!) |
tmux send-keys -t "$SESS" "<slash>" Enter
sleep 4
โ ๏ธ NIENTE Esc preventivo prima del slash command: rompe Kimi (interpreta /usage come testo) e per gli altri รจ inutile in sessione fresh.
Step 5 โ Capture pane
tmux capture-pane -t "$SESS" -p -S -100
Leggi il pane coi tuoi occhi LLM, estrai usage, reset, weekly.
Format Claude
Current session
โโโโโโโโ 45% used
Resets 8:10pm (UTC)
Current week (all models)
โโโโ 25% used
Resets Apr 27, 5am (UTC)
โ usage=45, reset=20:10 UTC, weekly=25
Format Kimi
โญโโโ API Usage โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ Weekly limit โโ 60% left (resets in 2d 27m) โ
โ 5h limit โโ 100% left (resets in 3h 27m) โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
โ usage=100-100=0, weekly=100-60=40
โ reset_at: ora_UTC + 3h 27m โ calcola HH:MM UTC
Format Codex
5h limit: [โโโโโโ] 35% left (resets 18:10)
Weekly limit: [โโโโโโ] 88% left (resets 14:00 on 02 May)
โ usage=100-35=65, weekly=100-88=12, reset=18:10
โ ๏ธ Codex riporta "% left", non % used: sottrai da 100.
Step 6 โ KILL sessione SUBITO
tmux kill-session -t "$SESS"
โ ๏ธ Killa SEMPRE anche se il parse fallisce. Niente sessioni persistenti.
Step 7 โ Scrivi sample nel JSONL
python3 /app/shared/skills/usage_record.py --manual \
--usage <X> --reset-at <HH:MM> --weekly <W> \
--provider <claude|kimi|openai> --source sentinella-worker
Step 8 โ Procedi con la decisione
Hai ora il sample fresco con proj/status calcolati da compute_metrics. Aggiorna memoria (vedi skill memory-state) e decidi ordine al Capitano (vedi skill order-formats).
Troubleshooting
Pane vuoto / nessuna modal
- CLI non ancora ready โ attendi altri 8s e ricaptura.
- Trust dialog ancora aperto โ invia
Enter (NON Escape: cancella e ti butta in bash).
Loading usage dataโฆ infinito (Claude)
TUI bloccato. Kill+respawn la sessione e riprova; se persiste cadi su L3 (FATAL).
Unrecognized command (codex)
Hai mandato /usage su codex. Manda /status e basta.
CLI morto / pane in bash
Kill la sessione e respawna.
