How-to

Manage storage & retention

You own the retention policy. Aarmos never rolls chats or receipts off-device unless you tell it to.

1
Settings → Storage & retention.
2
Review per-store space usage (chats, audit ledger, cache, vault).
3
Tune retention (default 30 days for chats, 90 for the signed audit ledger).
4
Purge on demand — hash chain re-anchors from the first surviving entry.
5
On the local runtime, evidence is committed durably: each append is written to a temporary file, flushed to disk, then renamed into place, and the containing directory is flushed too. A full disk or a read-only mount fails loudly with exit code 16 (evidence-durability) and leaves already-committed evidence intact, rather than truncating the chain.
6
Writes are coalesced by default. If your environment can lose power or be killed mid-run and you want every single append committed before the runtime continues, turn on synchronous commits — at some throughput cost.
```
AARMOS_SYNC_APPENDS=true aarmos proxy

# or in .aarmos/config.json: { "syncAppends": true }
```
7
Check that the workspace is durable and exclusive before you rely on it — aarmos doctor probes this and reports it above any connectivity symptom.
```
aarmos doctor
```