Settings, scheduling & the Agency Dashboard

Once you're running more than a couple of clients, the daily question stops being "what's wrong with this site?" and becomes "where do I spend the next thirty minutes?" Crawl Cove answers that with three connected pieces: a Settings page that configures the machine once, a Calendar + scheduler that makes the retainer rhythm run itself while the app is open, and an Agency Dashboard that triages the whole portfolio at a glance.

Settings: configure the machine once

The agency Settings page holds per-machine configuration that applies across every client:

• Agency default branding — your own logo and colours, the fallback when a client doesn't have its own.
• Data storage path with backup/export. The live database lives at <userData>/crawlcove.db; encrypted secrets sit separately under <userData>/secrets/. A one-click Back up database button writes a consistent snapshot of the whole SQLite file to a location you choose — it flushes the write-ahead log first, so the copy is safe to take while the app is running.
• Crawl defaults — concurrency, politeness delay, and user-agent string. These actually reach the crawler; set them once and every new crawl inherits them (still overridable per crawl).
• API keys — CrUX, Bing, and Open PageRank, all encrypted via safeStorage.
• LLM backend selector — see Configuring the optional LLM backend.

Calendar & the scheduler

The Calendar is colour-coded per client and holds scheduled crawls, report deadlines, client calls, and content due dates — the retainer rhythm made visible: monthly crawl → findings review → report → call.

• Natural-language quick add. Type something like a plain-English event and the date is resolved deterministically (by chrono-node); the model only handles the title and intent, and the feature degrades gracefully if the LLM is off. The date is never an AI guess.
• The scheduler runs due crawls while the app is open. A main-process scheduler fires the crawls you've scheduled — set a schedule in the New Crawl config when you run an audit. It runs while Crawl Cove is open rather than as a hidden background service, which keeps the local-first model honest: no daemon quietly touching client sites when you're not there.

The Agency Dashboard: 30-second triage

The Agency Dashboard is the global landing page — a thirty-second read on the whole book of business before you decide where the day goes.

It pulls together:

• A client health-score grid — one card per client, with a score and a trend arrow.
• An alert feed — new critical findings since yesterday, GSC traffic drops beyond a threshold, expired OAuth tokens, and lost high-value backlinks.
• Crawls currently running, with live page counts.
• Tasks due this week, across all clients.
• A "wins this month" counter — the number of verified-fixed findings, so the proof-of-value is sitting on your home screen.

The point is triage: glance, spot the two clients that need you today, ignore the twelve that are fine.

No silent failures: Connections health & the Activity Log

A multi-client tool is only trustworthy if you can tell when an integration breaks — so silent failure is designed out.

• Connections (per client) shows each integration's health: GSC OAuth status (token valid? last sync?), the Bing WMT key, GSC Links CSV import status, last-sync timestamps, recent sync errors, and Reconnect / Test buttons. When any chart in the app is empty, this page tells you why in one look.
• The Activity Log (agency level) is a chronological, filterable record of crawl events, tool calls with their actual arguments, LLM requests and responses, and errors with context — filterable by client and subsystem. When something misbehaves, the evidence is already on screen. (Secrets, query text, and robots bodies are never logged.)

Next

• Exporting your data: CSV & database backup — the full detail on the one-click database backup and the CSV exports.
• A tour of the interface — how the agency level and the per-client workspace fit together.