Dashboard

Dashboard

The Dashboard is the default landing view of every window. It answers four questions at a glance: Is Hermes running? What's it cost me? What did it just do? Is anything broken?

Layout

A scrolling stack of cards, refreshed automatically when ~/.hermes/state.db, config.yaml, or the logs change (HermesFileWatcher handles the reload).

Card	Source	What it shows
Hermes Process	pgrep (local) or ssh host pgrep (remote)	Running / stopped, PID, uptime. Start/stop/restart buttons.
Token Usage	state.db aggregations across all sessions	Input + output + cache + reasoning tokens; period selector (7/30/90 days / all).
Cost Tracking	Per-session actual_cost_usd (v0.7+) or estimated_cost_usd	Total spend in the period, broken down by model when available.
Recent Sessions	sessions table, ordered by started_at DESC	Last N sessions with title, source platform, message count, duration. Click to drill into Sessions.
Active Platforms	Parsed from config.yaml	Which messaging platforms are configured and their connection status.
Status	lastOpenError from HermesDataService	Surfaced when the database can't be read (missing sqlite3 on remote, permission denied, etc.). Links to Health for diagnostics.

Live refresh

Local windows watch ~/.hermes/ with FSEvents (DispatchSourceFileSystemObject). Remote windows poll mtimes every 3 seconds over the SSH ControlMaster. Either way, the Dashboard updates without needing a manual refresh — open it on a second monitor and watch sessions tick by as Hermes works.

Status pill in the toolbar

Every Scarf window has a connection pill in the toolbar showing the bound server and its state:

• Local — green dot, "Local" label.
• Remote (healthy) — green dot, server name.
• Remote (degraded) — yellow dot, "Can't read Hermes state" — opens a diagnostics sheet that runs 14 checks in one SSH session (connectivity, sqlite3 presence, read access on config.yaml/state.db, effective non-login $PATH).
• Remote (failed) — red dot, error message.

When something looks wrong

• "Hermes not running" — start Hermes from a terminal, or click the Start button.
• Empty cards — ~/.hermes/state.db is missing or unreadable. See First Run.
• Token/cost are zero but you've used Hermes — the schema may predate v0.7. Update Hermes; Scarf detects the new columns automatically.
• Yellow "Can't read Hermes state" pill on remote — open Manage Servers → Run Diagnostics. Each failed check explains why and how to fix.

Related pages

• Insights & Activity for deeper analytics.
• Chat for talking to the running Hermes.
• Servers & Remote for adding remote hosts and the diagnostics flow.

---

Last updated: 2026-04-20 — Scarf v2.0.1