Deployment intake
Each deployment is limited to 100 files and 50 MiB by default (operator hard caps 500 files / 100 MiB). Small deploys may use a single request; larger deploys use multi-request upload (init → file batches → finalize), with multi-file batches capped at 40 files or 8 MiB (a single large file may use the full deployment ceiling). Paths, content types, byte counts, and SHA-256 digests are validated before storage. Uploading creates an immutable ready candidate; traffic changes only on explicit release. Limit and quota failures return structured error codes with actionable CLI hints.
Static hosting rules
Optional _redirects and _headers files in the deploy root are parsed at finalize and stored with the deployment. Redirects support exact paths and trailing splats (/*), status codes 200 (rewrite), 301, 302, 303, 307, and 308. Status 200 targets must be relative. Headers support an allowlisted set (for example CSP, X-Frame-Options, Referrer-Policy); cookie and hop-by-hop headers are rejected. First matching redirect wins.
Files and data
Runtime file uploads currently use a JSON and base64 transport and are limited to 2 MiB per file. Runtime records are JSON documents. Service credentials are required for private data and file operations.
Realtime and AI
Realtime currently supports authenticated publishing and history polling; persistent live connections are not yet available. AI requests have bounded input and output sizes and may be rejected when a workspace budget or platform safety limit is reached.
Workspaces and credentials
A workspace can have up to 100 members and 25 teams. Invitations expire after seven days. Credentials can be restricted by app, environment, scope, origin, and expiry, and can be revoked at any time.
Fair use and changes
Deploylet may apply lower account or workspace limits to protect reliability and prevent abuse. Material limit changes are recorded in the changelog. Contact hello@deploylet.com before workloads that need higher limits.