> ## Documentation Index > Fetch the complete documentation index at: https://docs.archal.ai/llms.txt > Use this file to discover all available pages before exploring further. # Renew, destroy, and control usage > Keep sandboxes alive only while useful work is running and clean them up reliably. ## Renew ```bash theme={null} archal session renew --ttl 1800 ``` Renewal returns a fresh expiry and refreshed environment connections. Scoped provider credentials can rotate, so update long-running clients from the renewal response. Read the returned `lease` object when present: * `expiresAt` is the current TTL boundary * `idleExpiresAt` is the inactivity boundary, when one applies * `hardExpiresAt` is the absolute lifetime boundary * `renewable` tells you whether another extension is allowed Renewal never extends beyond the returned `hardExpiresAt`. Current API and web facade TTL enforcement are not yet uniform, so do not infer one plan-wide maximum. Treat the lease returned for the sandbox as authoritative. ## Destroy ```bash theme={null} archal session destroy ``` The default CLI command waits for confirmed destruction. Use `--detach` when a cleanup path needs fast durable acceptance, then wait separately if exact physical deletion is required. ```bash theme={null} archal session destroy --detach archal session wait --until destroyed ``` ## Usage Billing begins separately for each environment when it becomes ready and ends when teardown is durably accepted, the environment expires, or policy stops it. Failed provisioning and cold-start time are free. Each active environment costs \$0.10 per minute, prorated by the billed second. A sandbox with GitHub, Slack, and Datadog therefore accrues three environment minutes for every ready minute. Check the authoritative workspace balance and reserved usage with: ```bash theme={null} archal usage ``` Destroy promptly, bound CI parallelism, and create only the environments the test needs.