Use cases

One copilot.
Every job your servers throw at you.

Every card below is the agent doing what it actually does: you type the ask, it works over SSH on the host you're connected to, and by default anything that changes state waits for your approval.

Developers

Ship and debug without leaving your flow

01 — 03
⏱

3am incident response

Paging you at 3am shouldn't mean 40 minutes of grep. The agent tails logs, correlates errors across the services on the box, and hands you a root cause with the fix queued for approval.

you: api is 500ing since 02:40 — find out why and fix it
⇪

Deploys with a health check

Pull the release, restart the service, hit the health endpoint — one ask. Every step that changes the box renders as the exact command, and waits for your click.

you: ship latest main to staging and verify /health
✎

Config surgery over SFTP

Edit nginx configs, env files, and unit files remotely in the built-in file browser — and have the agent run the config test before you approve the reload.

you: bump client_max_body_size to 50M and reload safely
DevOps & SRE

Every box in the fleet, one session at a time

04 — 06
⛭

Provision new servers

Fresh VM to production-ready: users, firewall, runtime, service units. The agent proposes each step; you approve the ones that matter.

you: set this box up for the node app — hardened, pm2, nginx
↻

Patch & update hygiene

Check kernel and package versions on the host you're connected to, stage the upgrades, and schedule the reboot — every step a command you read first, not a blind apt on prod.

you: what needs patching on this box — start with openssl
◔

Resource triage

CPU spike, memory leak, disk at 89% — the agent finds the culprit process or directory and proposes the cleanup for your approval. The genuinely destructive shapes are denylisted: they ask every time, even with auto-approve on.

you: disk is nearly full — free space without touching app data
Security

Audit like you mean it

07 — 09
⛨

SSH & access audits

Review sshd_config, authorized_keys, and sudoers on any host. The agent flags password auth, stale keys, and over-broad sudo — and drafts the hardening change.

you: audit ssh access on prod — who can get in and how?
◉

Exposure checks

Open ports, listening services, firewall rules, TLS versions — a full surface check in one conversation, with fixes queued behind the approval gate.

you: what's listening on this box that shouldn't be public?
✓

Cert expiry checks

Check certificate expiry on a host before the browser warning does it for you — then leave a watcher re-running the check on a schedule while the app is open. Renewals are proposed, never auto-fired.

you: any TLS certs on this host expiring in the next 30 days?
Data & platform

Keep the stateful stuff healthy

10 — 12
▤

Database firefighting

Slow queries, lock contention, replication lag — the agent runs the pg_stat and EXPLAIN queries, reads the output, and tells you what's blocking what. Terminating a backend is its own approval.

you: the API is slow — check postgres for blocking queries
◫

Backup verification

Backups that were never restored are hopes, not backups. Check the cron ran, the dumps exist, and the sizes look sane — on the host you're connected to, in one ask.

you: did last night's db backup actually run on this host?
✱

Memory that survives the session

The copilot keeps persistent memory per server — services, quirks, past incidents — in a SQLite file on your Mac. Come back a month later and ask what it already knows about this host.

you: what do we know about this host and its incidents?

Your next incident deserves backup.

Download the app, add a host, and hand off the grunt work. 200 credits to start.