Limits & errors

Every number the platform enforces, in one place, so you never have to hunt for a cap in a guide.

Runtime

WhatLimit
Wake from sleepAbout a second, automatic on any message, trigger, or request
Chat reply budget (custom agents)Reply within 30 seconds; run longer work in the background and deliver it on the next message
exec (run a command)60s default timeout, 120s max; output is one merged stream capped at 256 KB
File transfers100 MB per file; any file call wakes a sleeping agent

Build & deploy

WhatLimit
Typical repo build + ship1 to 3 minutes (an idle builder cold-starts first, adding about a minute)
Silent build killA build that prints nothing for 5 minutes is killed
ZIP upload500 MB compressed, 2 GB unpacked
Template captureUsually 1 to 10 minutes (snapshots the image and volume)
RegistriesPre-built images from public registries only; private code goes through repo builds

Resources & plans

WhatLimit
Base allocation per agent2 GB RAM, 5 GB SSD
Self-serve ceilings per agent8 GB RAM, 100 GB SSD (flat add-on prices)
Included agentsFree 3 · Starter 20 · Growth 100 · Scale 500 (extras are flat per-agent prices)

Sizing details in Env vars & sizing, prices in Billing & plans.

HTTP errors

StatusMeaning
401Missing, invalid, revoked, or expired API key
402Plan limit reached, or an add-on needs a paid plan
403API key lacks the required scope for this endpoint
404No such agent, or it belongs to a different account
409Conflict: agent name taken, deploy already in progress, or a file-move destination exists
429Rate limited; back off and retry

The SDKs raise these as typed errors (SDK error classes); the CLI maps them to exit codes (CLI exit codes).