CloudAlly - Claude Code Skill

Prerequisites: Install the CLI

This skill drives the cloudally-cli binary. You must verify the CLI is installed before invoking any command from this skill. If it is missing, install it first:

macOS / Linux:

bash <(curl -fsSL https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cloudally/install.sh)

Windows (PowerShell):

iwr -useb https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/cloudally/install.ps1 | iex

Verify: cloudally-cli --version
Ensure ~/.local/bin (macOS / Linux) or %LOCALAPPDATA%\Programs\msp-skills (Windows) is on $PATH.

The installer downloads the cloudally-cli and cloudally-mcp binaries into ~/.local/bin (macOS / Linux) or %LOCALAPPDATA%\Programs\msp-skills (Windows). It does not register the skill with your agent and writes no MCP client config - see mcp-install.md for that wire-up.

If --version reports "command not found" after install, the runtime cannot see the binary directory on $PATH. Do not proceed with skill commands until verification succeeds.

Log in as one company, or as a partner that reaches every customer, and drive CloudAlly backups from scripts or an agent. The restore group plans, submits, watches and remembers multi-mailbox restores to a chosen backup date. The download group builds year-end mailbox archives for every mailbox backed up on one date and collects them to disk. backup snapshots reads a mailbox's past backups from the customer portal, which the documented API cannot, and backup health sweeps every customer for stale or failing backups.

When to Use This CLI

Use this CLI for anything that touches one CloudAlly company account (the default login) or, with a partner login, every customer account: listing backup tasks, restoring mailboxes and sites from a backup date, downloading mailbox archives as they stood on a past date, managing tasks and entities, running, pausing and cancelling backups, checking backup health across customers, managing accounts, resellers and users, and reading bills. It is the right tool when more than one mailbox needs restoring to the same date, when that date is in the past and each mailbox has to be resolved to the backup taken on or before it, or when restore history older than a week is needed.

Anti-triggers

Do not use this CLI for:

Do not use it to restore individual emails or files inside a mailbox by picking items in a browser; it restores whole entities (mailbox, site, drive) or JSON-specified item lists only.
Do not use it for Microsoft 365 administration itself (creating users, licenses, mailbox settings); it only drives CloudAlly backups of those tenants.
Do not use it to create API keys; client-id and client-secret are created in the CloudAlly partner portal.
Do not use it for billing reconciliation on indirect partner accounts; CloudAlly blocks billing endpoints by default for them.

Unique Capabilities

Batch restore by date

restore plan - Preview every mailbox-and-date restore a batch would submit, with the exact backup each mailbox will be restored from.

Reach for this before any multi-mailbox restore so the operator can review what will be restored and from which backup.
```
cloudally-cli restore plan --task 64b0e9 --date 2026-10-01 --date 2026-10-03 --agent
```
restore submit - Submit one restore per mailbox per date with a local ledger and a restore guard (restore-guard.db in the state directory) shared by every login and token that uses that state directory: a restore already sent from it is never sent again unless you pass --new-batch. Another OS user, --home, CLOUDALLY_HOME or CLOUDALLY_STATE_DIR has its own guard, and recovery restore and the MCP cloudally_execute tool bypass it; --resume keeps the batch's recorded target account and restore options.

Use this for every restore, a single mailbox included (--entities <mailbox>); recovery restore skips the plan, ledger, guard and confirmation.
```
cloudally-cli restore submit --task 64b0e9 --date 2026-10-01 --dry-run
```
restore watch - Follow a submitted batch until every restore finishes, with a per-mailbox tally and a non-zero exit if any failed.

Pick this to track a batch you submitted; use tasks list --type RESTORE only for raw job listings.
```
cloudally-cli restore watch --batch 20261015-120000-ab12 --interval 30s --agent
```
restore history - See what was restored for which company and when, from the local ledger, without calling the API.

Use this for audit questions about past restores; it never calls the API.
```
cloudally-cli restore history --since 2026-09-01 --agent
```
backup snapshots - List every past backup run for one mailbox in a given month, with each run's date in both ISO and the raw yyyyMMddHHmmss form a restore body needs.

Use this to confirm a past restore point exists before restoring anything. It is read-only and needs a partner login plus --account.
```
cloudally-cli backup snapshots --account 5f1c2a --task 64b0e9 --entity jane.doe@contoso.com --month 2025-12 --agent
```

Mailbox downloads to a past date

download submit - Create one download (export) job per mailbox and sub-source for every mailbox backed up on one date, record every row in a local ledger, and reconcile it against the portal.

Use this for year-end or legal-hold archives of many mailboxes, and for any MS365 mailbox. Pass --utc-offset so --on-date means the account's own calendar day, and run it once without --yes first: that prints the plan and creates nothing (exit 2; exit 0 with nothing to create when every mailbox was skipped, so read the skipped list before calling the set complete).
```
cloudally-cli download submit --account 5f1c2a --task 64b0e9 --on-date 2025-12-31 --utc-offset -5
```
download fetch - Download every ready archive to disk, one folder per year, mailbox and sub-source, minting and streaming each 60-second link in one step and resuming safely after an interruption.

Use this to collect the archives download submit created; --plan-only lists what it would write. --dest must already exist, even with --plan-only (mkdir archives; PowerShell: New-Item -ItemType Directory archives).
```
cloudally-cli download fetch --account 5f1c2a --dest ./archives --year 2025 --batch 20261015-201500-ab12 --plan-only
```
download start - Build one archive of one mailbox from the backup taken on or before a date, for single-archive sources (MS365 tasks go through download submit). It runs as a one-archive batch through the download guard, so running it again creates nothing; confirm the job with the printed download submit --batch <key> --resume --yes.
```
cloudally-cli download start --account 5f1c2a --task 64b0e9 --entity jane.doe@contoso.com --date 2025-12-31 --dry-run
```
download list, download watch and download link - List an account's download jobs, follow one to READY_FOR_DOWNLOAD, and mint a fresh link for a finished one. A link is valid for about 60 seconds: hand it to a program, never to a ticket.
```
cloudally-cli download watch --account 5f1c2a --job 9f2b1c --once
```

Partner operations

backup health - Find every mailbox or site across all customer accounts that has not backed up in N days or whose last attempt failed, with the error text.

Use this for a partner-wide health check instead of paging through distributor-accounts backup-statuses.
```
cloudally-cli backup health --stale-days 2 --group-by account --agent
```

Command Reference

distributor - Manage distributor

cloudally-cli distributor bills - bills
cloudally-cli distributor create-user - Create a secondary partner under a specific reseller.
cloudally-cli distributor delete-user - Delete a secondary user from a partner.
cloudally-cli distributor details - Returns partner details

distributor_account_tasks - Manage distributor account tasks

cloudally-cli distributor-account-tasks activate - Activate a backup task to run on schedule, if it was paused.
cloudally-cli distributor-account-tasks activate-all - Activate all accounts in a task for a backup.
cloudally-cli distributor-account-tasks activate-entity - Activate a specific account in a task for a backup.
cloudally-cli distributor-account-tasks add-entity - Manually add a site or account for a backup in a backup task.
cloudally-cli distributor-account-tasks backup-entity-now - Run a backup for the specified account or site.
cloudally-cli distributor-account-tasks backup-now - Run a backup task with all the activated accounts or sites in it.
cloudally-cli distributor-account-tasks cancel - Cancel the specified running task (an entire backup, restore or download tasks).
cloudally-cli distributor-account-tasks cancel-entity - Cancel the specified running task of a specified account or a site.
cloudally-cli distributor-account-tasks delete - Delete a backup task, removing all the data.
cloudally-cli distributor-account-tasks get - Get information for the specified task
cloudally-cli distributor-account-tasks list - List all tasks of a specified type: backup, restore, download (paginated).
cloudally-cli distributor-account-tasks pause - Pause execution by schedule for all accounts or sites of a backup task.
cloudally-cli distributor-account-tasks pause-entity - Pause execution by schedule of a specified account or a site.
cloudally-cli distributor-account-tasks remove-entity - Deactivate an account or site from a backup, removing all the coorresponding data.

distributor_accounts - Manage distributor accounts

cloudally-cli distributor-accounts access-token - Access token
cloudally-cli distributor-accounts backup-statuses - Get all active backup statuses.
cloudally-cli distributor-accounts create - Create a Cloudally account (trial status).
cloudally-cli distributor-accounts list - Get all users belonging to the partner.
cloudally-cli distributor-accounts sign-in-token - Sign in token
cloudally-cli distributor-accounts tasks - tasks
cloudally-cli distributor-accounts unsubscribe - Cancel subscription and delete a Cloudally account.

distributor_recovery - Manage distributor recovery

cloudally-cli distributor-recovery export - The documented export call; on the live API it answers 200 and creates no job. Use download submit or download start.
cloudally-cli distributor-recovery restore - Create a restore job for one entity (mailbox, site, drive) of a backup task from a specific backup date. Writes back into the customer's live mailbox, site or drive.

distributor_resellers - Manage distributor resellers

cloudally-cli distributor-resellers accounts - accounts
cloudally-cli distributor-resellers all-accounts - All accounts
cloudally-cli distributor-resellers all-backup-statuses - Get backup status for all resellers
cloudally-cli distributor-resellers all-bills - Get bills for all resellers
cloudally-cli distributor-resellers all-tasks - Get backup tasks for all resellers
cloudally-cli distributor-resellers backup-statuses - Get all active backup statuses.
cloudally-cli distributor-resellers bills - bills
cloudally-cli distributor-resellers create - Create a reseller
cloudally-cli distributor-resellers create-account - Create a Cloudally account (trial status).
cloudally-cli distributor-resellers create-user - Create a secondary user under a specific reseller.
cloudally-cli distributor-resellers delete-user - Delete a secondary user from a reseller.
cloudally-cli distributor-resellers get - Returns reseller by partnerId
cloudally-cli distributor-resellers list - Returns list of current distributor's resellers
cloudally-cli distributor-resellers tasks - tasks
cloudally-cli distributor-resellers unsubscribe-account - Cancel subscription and delete a Cloudally account.
cloudally-cli distributor-resellers update - update a reseller

recovery - Manage recovery

cloudally-cli recovery export - The documented export call; on the live API it answers 200 and creates no job. Use download submit or download start.
cloudally-cli recovery restore - Create a restore job for one entity (mailbox, site, drive) of a backup task from a specific backup date. Writes back into the customer's live mailbox, site or drive.

reseller - Manage reseller

cloudally-cli reseller bills - bills
cloudally-cli reseller create-user - Create a secondary user under the reseller.
cloudally-cli reseller delete-user - Delete a secondary user from a reseller.
cloudally-cli reseller details - Returns partner details

reseller_account_tasks - Manage reseller account tasks

cloudally-cli reseller-account-tasks activate - Activate a backup task to run on schedule, if it was paused.
cloudally-cli reseller-account-tasks activate-all - Activate all accounts in a task for a backup.
cloudally-cli reseller-account-tasks activate-entity - Activate a specific account in a task for a backup.
cloudally-cli reseller-account-tasks add-entity - Manually add a site or account for a backup in a backup task.
cloudally-cli reseller-account-tasks backup-entity-now - Run a backup for the specified account or site.
cloudally-cli reseller-account-tasks backup-now - Run a backup task with all the activated accounts or sites in it.
cloudally-cli reseller-account-tasks cancel - Cancel the specified running task (an entire backup, restore or download tasks).
cloudally-cli reseller-account-tasks cancel-entity - Cancel the specified running task of a specified account or a site.
cloudally-cli reseller-account-tasks delete - Delete a backup task, removing all the data.
cloudally-cli reseller-account-tasks get - Get information for the specified task
cloudally-cli reseller-account-tasks list - List all tasks of a specified type: backup, restore, download (paginated).
cloudally-cli reseller-account-tasks pause - Pause execution by schedule for all accounts or sites of a backup task.
cloudally-cli reseller-account-tasks pause-entity - Pause execution by schedule of a specified account or a site.
cloudally-cli reseller-account-tasks remove-entity - Deactivate an account or site from a backup, removing all the coorresponding data.

reseller_accounts - Manage reseller accounts

cloudally-cli reseller-accounts access-token - Access token
cloudally-cli reseller-accounts backup-statuses - Get all active backup statuses.
cloudally-cli reseller-accounts create - Create a Cloudally account (trial status).
cloudally-cli reseller-accounts list - Get all users belonging to the partner.
cloudally-cli reseller-accounts sign-in-token - Sign in token
cloudally-cli reseller-accounts tasks - tasks
cloudally-cli reseller-accounts unsubscribe - Cancel subscription and delete a Cloudally account.

reseller_recovery - Manage reseller recovery

cloudally-cli reseller-recovery export - The documented export call; on the live API it answers 200 and creates no job. Use download submit or download start.
cloudally-cli reseller-recovery restore - Create a restore job for one entity (mailbox, site, drive) of a backup task from a specific backup date. Writes back into the customer's live mailbox, site or drive.

session - Manage session

cloudally-cli session login - Returns the access and refresh tokens.
cloudally-cli session partner-login - Returns the access and refresh tokens.
cloudally-cli session refresh - Returns the access and refresh tokens.

tasks - Manage tasks

cloudally-cli tasks activate - Activate a backup task to run on schedule, if it was paused.
cloudally-cli tasks activate-all - Activate all accounts in a task for a backup.
cloudally-cli tasks activate-entity - Activate a specific account in a task for a backup.
cloudally-cli tasks add-entity - Manually add a site or account for a backup in a backup task.
cloudally-cli tasks backup-entity-now - Run a backup for the specified account or site.
cloudally-cli tasks backup-now - Run a backup task with all the activated accounts or sites in it.
cloudally-cli tasks cancel - Cancel the specified running task (an entire backup, restore or download tasks).
cloudally-cli tasks cancel-entity - Cancel the specified running task of a specified account or a site.
cloudally-cli tasks delete - Delete a backup task, removing all the data.
cloudally-cli tasks get - Get information for the specified task
cloudally-cli tasks list - List all tasks of a specified type: backup, restore, download (paginated).
cloudally-cli tasks pause - Pause execution by schedule for all accounts or sites of a backup task.
cloudally-cli tasks pause-entity - Pause execution by schedule of a specified account or a site.
cloudally-cli tasks remove-entity - Deactivate an account or site from a backup, removing all the coorresponding data.

user - Manage user

cloudally-cli user bills - Get current billing forecast.
cloudally-cli user create-primary - Create a Cloudally account (trial status).
cloudally-cli user create-secondary - Create a Cloudally user in a Cloudally account.
cloudally-cli user delete - Cancel subscription and delete a Cloudally account.
cloudally-cli user me - Get Cloudally Account settings.
cloudally-cli user secondaries - Get list of the Cloudally users of a Cloudally account.
cloudally-cli user status - Get status for all tasks of user.
cloudally-cli user update-password - Update password
cloudally-cli user update-permissions - Update permission setting of a Cloudally account user.
cloudally-cli user update-settings - Update Cloudally account settings.

backup

cloudally-cli backup health - Find every mailbox or site across all customer accounts that has not backed up in N days or whose last attempt failed, with the error text.
cloudally-cli backup snapshots - List every past backup run for one mailbox in a given month, with each run's date in both ISO and the raw yyyyMMddHHmmss form a restore body needs.

restore

cloudally-cli restore plan - Preview every mailbox-and-date restore a batch would submit, with the exact backup each mailbox will be restored from.
cloudally-cli restore submit - Submit one restore per mailbox per date with a local ledger and a restore guard (restore-guard.db in the state directory) shared by every login and token that uses that state directory: a restore already sent from it is never sent again unless you pass --new-batch. Another OS user, --home, CLOUDALLY_HOME or CLOUDALLY_STATE_DIR has its own guard, and recovery restore and the MCP cloudally_execute tool bypass it; --resume keeps the batch's recorded target account and restore options.
cloudally-cli restore watch - Follow a submitted batch until every restore finishes, with a per-mailbox tally and a non-zero exit if any failed.
cloudally-cli restore history - See what was restored for which company and when, from the local ledger, without calling the API.

download

cloudally-cli download submit - Create one download (export) job per mailbox and sub-source for every mailbox backed up on one date, with a local ledger and a reconcile pass.
cloudally-cli download fetch - Download every ready archive to disk, one folder per year, mailbox and sub-source.
cloudally-cli download start - Build one archive of one mailbox (single-archive sources; MS365 goes through download submit).
cloudally-cli download list - List an account's download (export) jobs.
cloudally-cli download watch - Follow one download job until its archive is ready.
cloudally-cli download link - Mint a fresh 60-second link for each file of a finished download job.

Finding the right command

When you know what you want to do but not which command does it, ask the CLI directly:

cloudally-cli which "<capability in your own words>"

which resolves a natural-language capability query to the best matching command from this CLI's curated feature index. Exit code 0 means at least one match; exit code 2 means no confident match - fall back to --help or use a narrower query. --json (and other machine formats) keep that exit-2 contract and write {"matches":[]} on stdout so agents can inspect the envelope without treating a miss as success.

Recipes

Plan a batch restore and save it for review

cloudally-cli restore plan --task 64b0e9 --date 2026-10-01 --plan-out plan.json

Resolve which backup each mailbox would be restored from and save the plan for review.

Submit the batch after reviewing the plan

cloudally-cli restore submit --task 64b0e9 --date 2026-10-01 --concurrency 3 --yes

Sends one restore per mailbox from the same plan restore plan showed (or pass --plan-file plan.json to submit a saved plan exactly). Without --yes nothing is sent.

Follow a batch to completion

cloudally-cli restore watch --batch 20261015-120000-ab12 --interval 30s --agent --select rows.entity,rows.status,rows.task_id

Polls the account's RESTORE tasks, correlates them to the batch and prints only the columns an agent needs.

Partner-wide health sweep (partner login)

cloudally-cli sync --resources distributor_accounts-backup-statuses && cloudally-cli backup health --stale-days 2 --group-by account

Windows PowerShell 5.1 has no &&; run the second command only when the first succeeded:

cloudally-cli sync --resources distributor_accounts-backup-statuses
if ($LASTEXITCODE -eq 0) {
  cloudally-cli backup health --stale-days 2 --group-by account
}

Needs auth login --partner; refreshes the statuses feed for every customer, then lists stale or failing entities grouped by customer.

What did we restore for a customer last quarter

cloudally-cli restore history --since 2026-07-01 --csv

Reads this login's local ledger only, so it works offline. The partner login, each auth use-account customer and each company login keep separate ledgers, so run it under the login that submitted the batches.

Check which backup a past date resolves to for one mailbox

cloudally-cli backup snapshots --account 5f1c2a --task 64b0e9 --entity jane.doe@contoso.com --month 2025-12

Lists that mailbox's backup runs for the month from the portal's backup calendar, so the restore point is confirmed before anything is restored.

Plan an end-of-year restore for every mailbox on a task

cloudally-cli restore plan --account 5f1c2a --task 64b0e9 --date 2025-12-31 --plan-out eoy.json

Resolves each mailbox to its newest successful backup at or before 2025-12-31 from the portal's backup calendar. Mailboxes with no backup by that date are listed as NO_BACKUP and left out of the submission.

Year-end archive of every mailbox on a task (partner login)

cloudally-cli download submit --account 5f1c2a --task 64b0e9 --on-date 2025-12-31 --utc-offset -5
cloudally-cli download submit --account 5f1c2a --task 64b0e9 --on-date 2025-12-31 --utc-offset -5 --yes

Creates one job per mailbox and sub-source for the backups taken on 31 December in the account's own timezone. A mailbox with no successful backup that day is listed as skipped, never archived from another date. Each row is saved as REQUESTED before its create is sent. A create whose result is unknown (timeout, dropped connection, server error) is UNKNOWN and is never sent again by --resume on its own: --resume first marks jobs it finds on the portal CREATED; add --retry-unknown only after download list shows they are missing. Rows that never appear on the portal are marked MISSING; re-send them with --batch <key> --resume --yes. The batch records --include-deleted, --include-legal-hold and --format, and --resume uses the recorded values (a different value is refused). Running the same submit again while its batch has archives created, possibly created or being sent is refused, with the --batch <key> --resume --yes command printed (a run with a different --utc-offset is a different submit, since it selects another day's backups); --new-batch starts a second batch and creates every export again, so never pass it on an agent's own initiative. The batch key and the resume command are printed before the first create. --timeout limits each API and portal call, not the whole run, so the reconcile wait runs to the end. Each create carries a unique job label (cloudally-cli <id>, visible in the portal's job list); a row is confirmed only by the job with its label (or, if the portal drops labels, by the one job of that snapshot created after the row was requested), never by an older export of the same snapshot, and a confirmed row stays created even after the job leaves the list. download submit claims every archive (account, task, mailbox, sub-source, snapshot, format, include flags) in one download guard (download-guard.db in the state directory) just before creating it, so no two batches or runs create the same archive; a zip and a PST, or the same snapshot with and without legal-hold items, are different archives. A new batch creates only the archives no other batch has created and lists the rest as SKIPPED (this batch's download fetch --batch downloads them through the holding batch's confirmed jobs), so a pilot mailbox followed by the whole task creates only the rest and one fetch collects all of it; it is refused only while another batch's create of one of its archives is still unsettled, with that batch's resume command. A row whose job is not found is EMPTY (an empty mailbox) only when this run sent it and the portal accepted it, the portal's job count grew by exactly the creates this run had accepted, and nothing else in this state directory (another download or restore, an earlier create of the batch not yet confirmed, or a create whose answer was lost) could have added a job; otherwise it is MISSING. Every --resume checks the portal before sending anything, --no-reconcile included (it skips only the wait after sending). When a confirmed job ends in failure (any status naming one: FAILED, CANCELLED, ABORTED, STOPPED, ERROR), download fetch (or the next resume) marks it JOB_FAILED and download submit --batch <key> --resume --yes creates that archive exactly once more. If a job of the same mailbox, sub-source and snapshot is listed but cannot be proved to be the batch's (its label was lost or edited), the row is held as ambiguous and never re-created: collect it with the printed download fetch --on-date ... --entities ... exactly as printed (its --on-date is the snapshot's UTC day, so do not add --utc-offset) and check it is the export wanted. While the batch folder (download-batches in the state directory) holds a ledger written by a pre-release build of these commands, one the download guard has no record of, download submit (with or without --resume, --dry-run included) and download start refuse before any request, exit 2 and name the file: check what it created with download list, then move the file away. download fetch --batch still reads it. The command exits 0 when every row is created, empty or held by another batch, 6 while rows are FAILED, MISSING, JOB_FAILED, UNKNOWN, ambiguous or REQUESTED and not yet confirmed, and 2 when it refused before sending anything; an interrupted reconcile wait exits 6 and prints the --batch <key> --resume --yes command, which confirms what was created and sends nothing that may already exist. Any other non-zero exit (1 when the portal session, ledger or guard could not be opened or saved, 3 or 4 from the task read) means stop and read the message; if a batch key was printed, continue only with download submit --batch <key> --resume --yes. A create the portal accepted whose job is still not listed an hour later becomes MISSING, and the next --resume sends it again. The job list never shows an export that finished empty, so if the same rows come back MISSING after a resend, stop resending, check download list, and leave the next step to a person. download submit --resume and download fetch --batch refuse (exit 2) an --account other than the batch's own. When the portal's job listing stops at its cap or cannot be read, nothing is concluded from a row's absence: no row becomes MISSING or EMPTY, and MISSING and UNKNOWN rows are not sent again in that run.

Collect the finished archives

cloudally-cli download fetch --account 5f1c2a --dest ./archives --year 2025 --batch 20261015-201500-ab12

Use the batch key download submit printed. Collect by --batch, not --on-date: fetch reads --on-date as the UTC day of the backup unless it is given the submit's --utc-offset, so without it fetch would miss the batch's archives or pick up another day's export. Create the folder first (mkdir archives; PowerShell: New-Item -ItemType Directory archives). Writes <dest>/2025/<mailbox>/<sub-source>/<archive>.zip with a small .cloudally.json record beside each archive. Re-runs skip an archive only when its record matches, so it can be re-run until every job is collected; existing files are never overwritten, and a name held by another file gets the export id added. Folders and files are readable only by the current user on macOS and Linux; on Windows they inherit the permissions of the --dest folder, so choose a private one. It exits 6 if any archive FAILED, is MISSING (including a body of the wrong size, an error page, or a link whose host resolves to a private address) or is UNLISTED (the portal's job listing stopped at its cap of 5,000 jobs, so the archive may still exist; nothing is suggested to create it), 8 when nothing failed but some jobs are still NOT_READY (run the same command later; a create still unconfirmed an hour after it was accepted, the batch's own or another batch's it holds, becomes MISSING with exit 6 and the resume command that settles it), and 3 when no download job matched at all. If the portal has already expired a confirmed archive, fetch says so and prints a --new-batch command; check download list first, and leave that decision to a person. With --batch it downloads only the jobs the batch confirmed, plus the archives other batches hold for it (through their confirmed jobs, matched by job id); confirm the rest with download submit --batch <key> --resume --yes; it never resends an UNKNOWN row, but a row the portal accepted over an hour ago and still does not list is MISSING and is sent again. download watch exits 0 when the job is READY_FOR_DOWNLOAD, 6 when it finished any other way (failed, cancelled, aborted, stopped or partly done), 3 when the job id is not on the account, and 8 with --once while it is still being built or when interrupted (Ctrl-C, or closing the terminal on macOS and Linux), printing the command to watch it again. Links to single-label hosts or .local, .internal, .lan, .home.arpa, .localhost and .corp names are refused. An archive streams as long as data keeps arriving; --timeout does not cut it off. .partial files left by a killed run are removed once they are an hour old. When several fetches run for the same customer, start them a few seconds apart.

Backup status across every customer (partner login)

cloudally-cli distributor-accounts backup-statuses --all --agent

One call returns the latest backup status of every active backup across the partner's customer accounts. Resellers use reseller-accounts backup-statuses; a distributor with resellers can use distributor-resellers all-backup-statuses.

List one customer's backup tasks

cloudally-cli auth use-account 5f1c2a
cloudally-cli tasks list --type BACKUP --all --agent --select id,source,status

auth use-account makes the customer's account-scoped token active, so tasks and user act inside that account. auth use-account --partner switches back.

Check the entities of one backup task

cloudally-cli tasks get 64b0e9 --agent --select entities.entity,entities.status

Each entity's lastBackupDates carries the latest backup date in the yyyyMMddHHmmss form that recovery restore --date takes.

Status of every task in the current account

cloudally-cli user status --all --agent

How CloudAlly behaves (read before restoring or downloading)

These were observed against a production CloudAlly partner tenant, not taken from the API reference. They decide which command to use.

The documented API knows only the latest backup. tasks get and the status feeds report each mailbox's most recent backup per sub-source. Every past backup date comes from the customer portal's backup calendar (getSnapshotsCalendar), which backup snapshots, restore plan, download start and download submit read for you. The portal is reached through the partner handoff: the partner token calls POST /v2/distributors/accounts/{id}/partner-access-token, which returns a single-use opaque token that is spent at app.cloudally.com/newapp?accessToken=. So any past-date read needs auth login --partner and --account <id>.
extid is mandatory for the calendar. Without the mailbox's external id the calendar answers success with no rows, which reads as "no backups". The CLI looks it up from the task; pass backup snapshots --ext-id only when the task record lacks it.
Snapshot timestamps are UTC. A nightly backup that runs late in the evening local time is stamped the next UTC day, so the run stamped 2022-12-31 for a US Eastern account holds the 30 December state. download submit --utc-offset -5 selects by the account's own calendar day; leaving it at 0 matches the UTC date, which is usually the day before the one intended.
The documented export endpoint does nothing. POST .../recovery/export (the generated distributor-recovery export, reseller-recovery export and recovery export commands) answers 200 with an empty body and creates no job. Downloads are created the way the portal creates them, through download start (single-archive sources) and download submit (MS365, any number of mailboxes).
An MS365 mailbox is one download job per sub-source. Mail, calendar, contacts, tasks and notes (O365MAIL, O365CALENDAR, O365CONTACTS, O365TASKS, O365NOTES) are separate jobs, each asked for with that sub-source's own timestamp. A job for the parent MS365 source is accepted and creates nothing, which is why download start refuses MS365 tasks and points at download submit.
Download links last 60 seconds. A link is a presigned object-storage URL. download fetch mints and streams each one in the same step. Never paste a link into a ticket or chat; it is dead within a minute, and until then anyone holding it can download the archive. Customers can collect finished downloads themselves in the CloudAlly portal (their account name > Jobs > Download results).
Finished download jobs cannot be deleted. tasks delete (and the distributor and reseller variants) is DELETE .../backup-tasks/{id}: it deletes a backup task together with its backed-up data. Never point it at a download job id.
A restore is sent once per state directory, and an agent never re-sends one. Restores write into live mailboxes. Run restore plan first (or restore submit without --yes, which lists each mailbox and the backup it resolves to and sends nothing; --dry-run without --plan-file prints placeholder bodies only), show the human the plan, and pass --yes only after they confirm the account, task, mailboxes and dates. restore submit claims each restore in a guard (restore-guard.db in the state directory, shared by every login and token that uses it) and refuses a run that would send one again. Another OS user, --home, CLOUDALLY_HOME or CLOUDALLY_STATE_DIR has its own guard, and recovery restore and the MCP cloudally_execute tool bypass it, so run restores from one account and one state directory. The guard also refuses any restore of the same task and mailbox whose resolved backup is within 30 minutes of one already sent (the same backup reported two ways). Ledgers written by pre-release builds are not imported: while the data directory holds a restore ledger with a batch the guard has no record of, restore submit refuses before any request, exits 2 and names the file. Check what it sent with restore history under the login that wrote it (a data.db predates per-user stores and does not show there, so treat its restores as sent), then move the file away; moving a data-<hash>.db also moves that login's synced rows and restore history with it. restore history and restore watch still read it. Never pass --new-batch on your own initiative: it restores those mailboxes a second time. If restore submit reports UNKNOWN rows, run only restore watch --batch <id> --once and show the human the result; the printed --resume --retry-unknown --yes command is theirs to run, and only for rows watch still finds no RESTORE task for. restore history shows the ledger of the login you are on (partner, each auth use-account customer and each company login keep their own). restore submit exits 0 only when every row was accepted, 1 when the ledger or guard cannot be read (nothing sent), 2 when it refused before sending, 3 when no mailbox has a backup at or before the date or the task id is not found, 4 on a 401/403, 10 when auth-context.json cannot be trusted (nothing sent), and 5 when any row failed, is UNKNOWN or was never sent (for example after Ctrl-C), printing the exact --resume command, or when reading the task failed before anything was sent; restore watch exits 0 only when every row was submitted and its RESTORE task reports success (COMPLETED, COMPLETE, SUCCEEDED, SUCCEED, SUCCESS, DONE or FINISHED); 5 means something needs a person (a restore failed, was cancelled or only partly succeeded, a task reports a status the CLI does not recognise (printed as received), a row is still UNKNOWN, a row was never submitted, or a restore was accepted but no RESTORE task ever appeared: check tasks list --type RESTORE and do not resend); 8 means restores are still running, scheduled or paused (--once, --max-wait or Ctrl-C), so run watch again later. A --retry-unknown resend that is refused leaves the row UNKNOWN, because the first send may still have been accepted. --timeout bounds each API or portal request in restore plan, submit and watch, never the whole run; restore watch --max-wait (default 2h) bounds the watch. An interrupted submit (Ctrl-C or a tool timeout) says how many restores were already accepted and prints the --resume command; never report it as nothing restored.
One portal session per customer at a time. Starting several portal sessions for the same customer in the same instant can invalidate each other's unspent handoff tokens (seen as portal session is not authenticated (status 401)). When running several download fetch or download submit processes for one customer, start them a few seconds apart (about 6 seconds was enough).

Auth Setup

CloudAlly needs three things on every request: a client-id header, a client-secret header (the API key pair created under Integrations > REST API in the partner portal) and a Bearer access token. The key pair identifies the integration; what the CLI can reach depends only on which login you use. Set CLOUDALLY_CLIENT_ID and CLOUDALLY_CLIENT_SECRET, then set CLOUDALLY_EMAIL and CLOUDALLY_PASSWORD to one company's CloudAlly account and run cloudally-cli auth login: that is the default, and every command is limited to that one company. For partner-wide work (backup health across customers, the distributor-* and reseller-* commands, switching between customers with auth use-account <accountId>), set CLOUDALLY_PARTNER_EMAIL and CLOUDALLY_PARTNER_PASSWORD and run cloudally-cli auth login --partner; that login can see and change every customer account. auth use-account --partner switches back. Tokens are stored in the credentials file and refreshed automatically. Past backup dates, downloads and download links come from the customer portal, reached through the partner handoff, so those commands need the partner login and --account <id>.

Run cloudally-cli doctor to verify setup.

Agent Mode

Add --agent to any command. Expands to: --json --compact --no-input --no-color.

Global format flags share one contract on promoted, novel, sync, and --deliver paths:

--json - one JSON document on stdout (sync progress events go to stderr)
--compact - keep identity/status/timestamp fields; does not change the document vs stream shape
--csv / --plain - tabular rows (collection envelopes unwrap to the row array)
--quiet - one identity value per row, no envelope
Pipeable - JSON on stdout, errors on stderr
Filterable - --select keeps a subset of fields. Dotted paths descend into nested structures; arrays traverse element-wise. Critical for keeping context small on verbose APIs:
```
cloudally-cli tasks list --agent --select id,source,status
```
Previewable - --dry-run shows the request without sending
Offline-friendly - sync/search commands can use the local SQLite store when available
Erasable local data - local-data purge previews the store files, response cache and learn journal it would delete and deletes only with --yes; it is not exposed over MCP. CLOUDALLY_NO_LOCAL_DATA=1 turns off the response cache and API-response store writes, turns auto reads into live reads, and blocks sync and workflow archive
Non-interactive - never prompts, every input is a flag
Explicit confirmation - --agent does not imply --yes; pass --yes separately only after the target, arguments, and side effects are clear
Explicit retries - use --idempotent only when an already-existing create should count as success, and use --ignore-missing only when a missing delete target should count as success

Response envelope

Commands that read from the local store or the API wrap output in a provenance envelope:

{
  "meta": {"source": "live" | "local", "synced_at": "...", "reason": "..."},
  "results": <data>
}

Parse .results for data and .meta.source to know whether it's live or local. A human-readable N results (live) summary is printed to stderr only when stdout is a terminal AND no machine-format flag (--json, --csv, --compact, --quiet, --plain, --select) is set - piped/agent consumers and explicit-format runs get pure JSON on stdout.

Paths and state

Agents should treat the CLI's path resolver as part of the runtime contract:

Use --home <dir> for one invocation, or set CLOUDALLY_HOME=<dir> to relocate all four path kinds under one root.
Use per-kind env vars only when a specific kind must diverge: CLOUDALLY_CONFIG_DIR, CLOUDALLY_DATA_DIR, CLOUDALLY_STATE_DIR, CLOUDALLY_CACHE_DIR.
Resolution order is per-kind env var, --home, CLOUDALLY_HOME, XDG (XDG_CONFIG_HOME, XDG_DATA_HOME, XDG_STATE_HOME, XDG_CACHE_HOME), then platform defaults.
config contains settings like config.toml and profiles. data contains credentials.toml, the local store (data-<hash>.db, one per CloudAlly user and API URL; data-nocredential.db for runs with no credential), cookies, and auth sidecars. state contains persisted queries, jobs, teach.log, restore-guard.db and download-guard.db (the restore and download guards, shared by every login and token that uses this state directory), download batch ledgers, and auth-context.json, which every auth login writes and which, after auth login --partner, also holds the partner access and refresh token; auth logout deletes it. cache contains regenerable HTTP/cache files.
Stored secrets live in credentials.toml under the data dir. Existing legacy config.toml secrets are read for compatibility and leave config.toml on the first auth write.
Run cloudally-cli doctor --fail-on warn to surface path and credential-location warnings. agent-context exposes a schema v4 paths block for agents that need the resolved dirs.

For MCP, pass relocation through the MCP host config. The MCP binary does not inherit CLI flags:

{
  "mcpServers": {
    "cloudally": {
      "command": "cloudally-mcp",
      "env": {
        "CLOUDALLY_HOME": "/srv/cloudally"
      }
    }
  }
}

Fleet precedence: an inherited per-kind env var overrides an explicit --home for that kind. Use CLOUDALLY_HOME or per-kind vars as durable fleet levers, and use --home only for a single invocation. Relocation is not reversible by unsetting env vars; move files manually before clearing CLOUDALLY_HOME, or doctor will not find credentials left under the former root.

Automatic learning

This CLI ships a self-capturing learning loop. The CLI does its own bookkeeping: every invocation is journaled locally, a failed flag followed by a corrected retry auto-derives a flag_alias candidate, and a teach on a query family without a playbook auto-synthesizes a playbook_candidate from the session's journal. Your job is judgment only: recall first, act on surfaced candidates, teach the final answer, playbook amend when you observe a correction. You never record failures by hand.

Step 1: `recall` before any discovery

Before list/search/drill commands on a new user question, pass the question as an argv or MCP tool argument to recall --agent. Do not interpolate user-controlled text into a shell command line.

Quoted recall "<question>" breaks on an apostrophe, which is ordinary English. A quoted heredoc breaks when a body line equals the delimiter, and that delimiter is published in these docs. Write the question with a non-shell file-writing tool, then read it back as data:

# Write the question verbatim with your file-writing tool (no shell involved).
# Command substitution on a file only ever yields data - the shell never
# parses the file's bytes as syntax.
QUERY=$(cat /path/to/question.txt)
cloudally-cli recall "$QUERY" --agent

Prefer MCP: pass the question as the tool's query argument. "$QUERY" after a file read is argv-safe; putting the question itself in the command text is not.

The response envelope:

{
  "query": "...",
  "normalized": "<normalized form>",
  "query_entities": ["..."],
  "found": true | false,
  "match_score": 0.0,
  "results": [
    { "resource_id": "...", "resource_type": "...", "venue": "...",
      "confidence": 2, "entity_match": "exact|partial|unknown",
      "source": "taught|preseed|pattern", "warnings": ["..."] }
  ],
  "mismatches": [ /* only when --debug-mismatches */ ],
  "warnings": [ /* top-level */ ],
  "candidates": [
    { "id": 12, "class": "flag_alias | playbook_candidate",
      "summary": "...", "sightings": 3, "last_seen": "...",
      "rationale": "...",
      "next_action": ["<trial command>", "cloudally-cli learnings confirm 12"] }
  ],
  "playbook": {
    "query_family": "...",
    "playbook": {
      "steps": [ { "cmd": "<command with {slot} substitution>", "purpose": "..." } ],
      "entity_slots": ["$ENTITY"],
      "expected_tool_calls": 3
    },
    "slots_resolved": { "$ENTITY": { "token": "<live token>", "canonical": "<canonical>" } },
    "notes": "<workarounds + gotchas for this query family>"
  },
  "notes": "<duplicate surface for non-playbook callers>"
}

Empty-store short-circuit: if the store has no learnings, playbooks, or candidates yet (recall finds nothing and learnings list and learnings candidates are both empty), skip recall for the rest of this session instead of taxing every query; resume recall-first once something has been taught.

Step 2: decision tree

Read candidates, playbook, notes, results[0], and warnings in that order:

if Candidates present (warnings include "candidates_present"):
    -> candidates are try-then-confirm, never facts. Follow each candidate's
       two-step next_action verbatim: run the trial command first, then run
       `learnings confirm <id>` only after the trial verified the behavior.
       Reject a wrong candidate with `learnings reject <id>`.
    -> NEVER re-teach something recall surfaced as a candidate; confirm or
       reject that candidate instead of teaching a duplicate.
    -> candidates ride alongside playbooks and resource hits, not instead of
       them; continue with the branches below after acting on them.

if Playbook present:
    -> READ Playbook.notes verbatim FIRST (workarounds + gotchas the CLI surface doesn't expose)
    -> replay Playbook.steps in order, substituting Playbook.slots_resolved entries
       for the entity slot tokens. If a step's slot is unresolved, fall back to
       discovery for that step only.
    -> the Playbook's expected_tool_calls is a budget; if you find yourself running
       materially more, record the divergence via `cloudally-cli playbook amend`
       at end-of-session.

elif Notes present (no Playbook):
    -> read Notes verbatim before any discovery step; they carry known gotchas
       for this query family even when no structured choreography exists yet.

elif Found AND Results[0].EntityMatch == "exact" AND Results[0].Confidence >= 2:
    -> skip discovery; fetch live data for Results[*].ResourceID in parallel

elif Found AND Results[0].EntityMatch == "partial":
    -> candidate hint, NOT a hit; read the resource title to validate before trusting

elif (any row in Mismatches[] when --debug-mismatches was passed):
    -> treat as cold start; the stored learning is for a different entity
       (different canonical resolved from query_entities)

else:  // Found == false, no playbook, no notes
    -> cold start; run discovery normally; teach the answer afterward (Step 4).
       If the family has no playbook yet, that teach auto-synthesizes a
       playbook candidate from this session's journal - you do not need to
       record one by hand.

Playbook and Notes are orthogonal to the per-resource path. A recall response can carry both a Playbook AND a Results[] hit - use both: the Playbook tells you which choreography to run; the resource hits short-circuit specific steps. Default to skipping mismatches; pass --debug-mismatches only when investigating cold-start surprises.

Candidate judgment details: learnings confirm <id> prints the candidate's full payload before materializing it - check that the printed payload matches the behavior you verified. learnings reject <id> tombstones the derivation signature so the same candidate does not resurface. The envelope carries only the few candidates worth acting on now; cloudally-cli learnings candidates lists the full open set.

Graceful degradation: if learnings confirm is an unknown command, you are driving an older binary - ignore the candidates guidance and follow the rest of the protocol.

Step 3: always read `warnings`

low_confidence: row exists at confidence<2. Treat as a hint, not a skip-discovery hit.
resource_not_in_store: the local store doesn't have the resource the learning points at. The match validator couldn't classify entities - direct-fetch and re-evaluate.
cross_alias_match (per-result): the row was taught under a different alias and matched the live query's canonical via entity_lookups (e.g., a "USA" teach satisfying a "United States" recall). Trust the resource_id.
similar_shape_different_entity:<canonical> (top-level): a structurally matching row exists but its canonical entity differs from the live query's. Treated as cold start; the warning carries the conflicting canonical as a hint, but the row is NOT promoted into Results.
ambiguous_alias (top-level): a single query entity resolved to multiple canonicals (e.g., "Cards" → Arizona Cardinals + St. Louis Cardinals). Surface the ambiguity from context before committing to a resource.
candidates_present (top-level): the envelope carries a candidates section. Handle it via the candidates branch in Step 2 before anything else.
lookup_refresh_available (top-level): an entity in the query has no lookup row yet, but synced data could provide one. Run cloudally-cli sync to refresh entity lookups.
Top-level no_learnings_for_query_family: the table had no rows above the Jaccard floor. Pure cold start.

Step 4: `teach &` after finalizing your response - always

Teaching is unconditional. After resolving a query the store could not answer, background-teach the final resource mapping - no call-count threshold, no judging whether it was "worth" learning. The teach is the anchor of the loop: it triggers playbook synthesis for a family without a playbook, and same-referent phrasings fold into one family so near-duplicate teaches do not fragment the store. Fire it after assembling your user-facing response but BEFORE emitting it, with a shell & so the call returns immediately. Pass the query the same way as recall - argv/MCP, or file-then-$QUERY. Do not splice the question into the command text:

QUERY=$(cat /path/to/question.txt)
cloudally-cli teach --query "$QUERY" --resource-type <type> --resource <id1> --resource <id2>
# (append shell `&` to background it)

Silent on success. Errors only land in teach.log under the resolved state dir. Teach the most specific resource - if the user asked a broad question and you walked through parent records to find the specific answer, teach the leaf id, not the parent. The CLI uses seeded entity_lookups for cross-alias resolution at recall time, so a teach under one alias (e.g., "Niners") satisfies future queries under another alias (e.g., "49ers", "San Francisco") automatically.

PII rule: teach the structural question with identifiers stripped - never include names, emails, phone numbers, account ids, or other personal identifiers in taught queries or notes. The CLI scans teach queries for obvious email/phone shapes and warns, but does not block; strip before teaching rather than relying on the warning.

Step 5: playbooks - optional flags, automatic synthesis

You do not need to decide whether a session "deserves" a playbook: a teach on a family without one auto-synthesizes a playbook_candidate from the session's journal, and the next session judges it via confirm/reject. Attach explicit playbook flags only when you already hold choreography worth recording verbatim - workarounds the CLI didn't surface (silently-dropped flags, undocumented params, pagination tricks, payload gotchas). Prefer the integrated one-call form - record the resource learning and the playbook in the same teach invocation:

# Common case: record both the resource learning AND the playbook in one call.
QUERY=$(cat /path/to/question.txt)
cloudally-cli teach \
  --query "$QUERY" \
  --resource <id> \
  --playbook-file ~/playbooks/<shape>.json \
  --playbook-notes-file ~/playbooks/<shape>-notes.md
# (append shell `&` to background it)

# Alternate: playbook-only (no resource to record alongside).
QUERY=$(cat /path/to/question.txt)
cloudally-cli teach-playbook \
  --query "$QUERY" \
  --playbook-file ~/playbooks/<shape>.json \
  --notes-file ~/playbooks/<shape>-notes.md

Playbook files are JSON with steps, entity_slots, expected_tool_calls. Notes files are markdown carrying the gotchas verbatim. File-free callers (MCP-only agents) pass the same content inline: --playbook-json and --playbook-notes on the integrated teach form, --playbook-json and --notes on teach-playbook. On the integrated teach form, the playbook flags are optional - omit them entirely for a resource-only teach. On the standalone teach-playbook form, at least one of the playbook and notes flags must be set; both empty is rejected. Playbooks are keyed on the structural query family (entities stripped) so a recipe taught from one entity-shaped query applies to every other query of the same shape, with slots_resolved binding the live query's canonical at recall time.

When you DO find a playbook on a future recall, treat it as ground truth: replay the steps with slots_resolved substitutions, skip the discovery that the choreography already documents, and read notes before any step.

Step 6: `playbook amend &` when your debug response identifies a correction

If your debug-protocol response identifies a concrete correction the notes or playbook should know - a workaround, an undocumented endpoint shape, a stale field name, observed schema drift, an empty-payload fallback - fire playbook amend BEFORE emitting your user-facing response. Same fire-and-forget posture as teach. Pass the query and note as argv/MCP arguments, or write each with a non-shell file tool and read them back (QUERY=$(cat ...), NOTE=$(cat ...)). Do not interpolate either string into the command text:

QUERY=$(cat /path/to/question.txt)
NOTE=$(cat /path/to/note.txt)
cloudally-cli playbook amend \
  --query "$QUERY" \
  --add-note "$NOTE"
# (append shell `&` to background it)

What counts as worth amending: a behavior you OBSERVED this session that future-you would benefit from knowing. Examples worth amending:

A workaround for a CLI surface that silently drops or misorders a flag.
An undocumented endpoint shape (response wrapped in {meta, results}, payload nested two levels deeper than the docs claim).
Observed schema drift (a field renamed, an index that shifted between seasons, a category label that the API now returns lower-cased).

What does NOT belong in notes:

The year-specific or entity-specific answer to the user's question. That's the response, not a learning.
Per-team / per-athlete / per-row data the playbook already retrieves at runtime.
Statements that paraphrase what the existing notes already say.

The amend command appends to the family's existing notes with a timestamped marker ([amend YYYY-MM-DDTHH:MMZ]: <text>). Multiple amends accumulate; the audit trail is visible. If no playbook exists yet for the family, amend creates a notes-only one (so cold-start corrections still land).

PII discipline for amend notes

playbook amend notes are designed to potentially flow upstream as shared knowledge in future versions of the Printing Press. Keep them clean of user-identifying content so the upstream-contribution path stays open without retroactive scrubbing:

Do NOT embed paths to user filesystems, personal API keys or tokens, user email addresses, user GitHub handles, or specific query histories tied to a single user.
Acceptable: endpoint shapes, undocumented field names, API gotchas, observed schema drift, workarounds for CLI surfaces, generalizable pagination or retry tactics.

If a correction is only meaningful with user-specific context, it belongs in a personal note, not in the playbook amend.

Measuring the loop

cloudally-cli learnings stats reports recall hit rate, teach-to-reuse, playbook resolution rate, and candidate confirm/reject counts from the local learn_events table. Rates are null until they have a denominator; everything stays on this machine. Use it to check whether the loop is earning its keep for this CLI.

Disabling learning

--no-learn on a single command short-circuits both recall and the teach write path. Use for deterministic agent flows or tests that must not be affected by accumulated learnings.
CLOUDALLY_NO_LEARN=true in the environment globally disables the pipeline.

Agent Feedback

When you (or the agent) notice something off about this CLI, record it:

cloudally-cli feedback "the --since flag is inclusive but docs say exclusive"
cloudally-cli feedback --stdin < notes.txt
cloudally-cli feedback list --json --limit 10

Entries are stored locally as feedback.jsonl under the resolved data dir. They are never POSTed unless CLOUDALLY_FEEDBACK_ENDPOINT is set AND either --send is passed or CLOUDALLY_FEEDBACK_AUTO_SEND=true. Default behavior is local-only.

Write what surprised you, not a bug report. Short, specific, one line: that is the part that compounds.

Output Delivery

Every command accepts --deliver <sink>. The output goes to the named sink in addition to (or instead of) stdout, so agents can route command results without hand-piping. Three sinks are supported:

Sink	Effect
`stdout`	Default; write to stdout only
`file:<path>`	Atomically write output to `<path>` (tmp + rename). Binary-response commands write decoded payload bytes (not the base64 JSON envelope) and print a small JSON receipt on stdout; `--json`/`--csv` do not refuse when this sink is set.
`webhook:<url>`	POST the output body to the URL (`application/json`)

Unknown schemes are refused with a structured error naming the supported set. Webhook failures return non-zero and log the URL + HTTP status on stderr.

Named Profiles

A profile is a saved set of flag values, reused across invocations. Use it when a scheduled or recurring agent reuses the same saved flags while providing different input each run.

cloudally-cli profile save briefing --json
cloudally-cli --profile briefing tasks list
cloudally-cli profile list --json
cloudally-cli profile show briefing
cloudally-cli profile delete briefing --yes

Explicit flags always win over profile values; profile values win over defaults. agent-context lists all available profiles under available_profiles so introspecting agents discover them at runtime.

Async Jobs

For endpoints that submit long-running work, the generator detects the submit-then-poll pattern (a job_id/task_id/operation_id field in the response plus a sibling status endpoint) and wires up three extra flags on the submitting command:

Flag	Purpose
`--wait`	Block until the job reaches a terminal status instead of returning the job ID immediately
`--wait-timeout`	Maximum wait duration (default 10m, 0 means no timeout)
`--wait-interval`	Initial poll interval (default 2s; grows with exponential backoff up to 30s)

Use async submission without --wait when you want to fire-and-forget; use --wait when you want one command to return the finished artifact.

If --wait times out or polling fails after the job was accepted, the command exits 8 and prints the job ID plus a recovery command that fetches the result. The job may still be running and may already be billed, so run the recovery command instead of resubmitting.

Exit Codes

Code	Meaning
0	Success
2	Usage error (wrong arguments)
3	Resource not found
4	Authentication required
5	API error; for `restore submit` and `restore watch` it means rows failed, are UNKNOWN or need a person: read the message and never re-run the original command
6	Partial failure
7	Rate limited (wait and retry)
8	Async job submitted but not finished (run the printed recovery command; do not resubmit)
10	Config error

Argument Parsing

Parse $ARGUMENTS:

Empty, help, or --help → show cloudally-cli --help output
Starts with install → ends with mcp → MCP installation; otherwise → see Prerequisites above
Anything else → Direct Use (execute as CLI command with --agent)

MCP Server Installation

Install the MCP binary (run the install script from the Prerequisites section, or see mcp-install.md for per-agent wire-up).

claude mcp add cloudally-mcp --env CLOUDALLY_CLIENT_ID="<your client id>" --env CLOUDALLY_CLIENT_SECRET="<your client secret>" -- cloudally-mcp

Run cloudally-cli auth login --partner (or auth login for one company) once in a terminal; the MCP server reads the saved token and refreshes it. Through MCP, restore_submit, download_submit and download_start only preview (the server exposes no --yes), and download fetch is CLI-only: review the plan, then submit from a terminal.
Verify: claude mcp list

Direct Use

Check if installed: cloudally-cli --version (which cloudally-cli on macOS and Linux, Get-Command cloudally-cli in PowerShell) If not found, offer to install (see Prerequisites at the top of this skill).
Match the user query to the best command from the Unique Capabilities and Command Reference above, and read "How CloudAlly behaves" before any restore or download.
Execute with the --agent flag:
```
cloudally-cli backup health --agent
```
If ambiguous, drill into subcommand help, for example cloudally-cli download --help.