Memory quotas explained: upload, storage, retrieve, and preview
How Engra maps R2 storage economics to monthly upload and atom storage caps, plus retrieve and source-preview limits that align with wake_up, search, and recall.
Four quota lines
Every plan tracks Memory usage in four buckets: • Upload — bytes ingested per month (documents and source archives). • Storage — cumulative atom markdown + attachments in object storage. • Retrieve — search, recall, and wake-up API calls per month. • Preview — presigned source-file open bytes per month.
The dashboard usage page shows current-period consumption against your plan limits.
Upload vs storage
Upload is ingress: new content entering your library this month. Storage is what remains indexed—the atom bodies and attachments that accumulate over time.
Licensed plans (Growth, Pro) derive upload and storage caps from an R2 cost envelope—a fraction of list price allocated to Memory infrastructure. Starter uses explicit caps suited for proof of concept.
Retrieve maps to agent calls
Each memory_search, memory_recall, and memory_wake_up counts as one retrieve request. Heavy IDE agents should install Agent Skills so wake_up runs once per session instead of on every message.
Starter includes 5K retrieve/mo. Growth includes 100K. Pro includes 500K. Enterprise is custom. PAYG meters every call with no included bundle.
Source preview
When agents open original source files through the API, preview bytes accrue separately from atom storage. Growth and Pro include unlimited source previews; Starter includes a monthly preview cap.
What happens at the cap
Starter returns quota-exceeded responses—upgrade or wait for the next billing period. Licensed plans and PAYG continue with metered overage billed through Stripe.
Compare plans on pricing or read choosing a plan for upgrade timing.