Documentation

Limits and quotas

The ceilings that apply to projects, tasks and execution, and what happens when you reach one.

Limits differ by plan. This page describes which limits exist and how they behave; the current numbers for each plan are on the pricing page.

Projects

The free plan allows a single project. Paid plans do not cap projects.

Agent slots

Slots cap how many tasks run at once, not how fast any task finishes. The free plan includes one low-priority slot; Professional includes one standard slot with more available per month; Enterprise runs on contracted capacity with a minimum commitment.

When every slot is busy, new tasks queue in PENDING and start as slots free up, ordered by priority. Nothing is rejected for lack of capacity — it waits.

See Capacity and agent slots.

Agent-hours

Plans with a monthly agent-hour allowance meter time while an agent is actively working. Queue time is free.

When the allowance is exhausted, new work stops starting until the allowance resets or you move to a plan with more. Work already running is not killed mid-task.

Maximum task run time

A single task may run for a bounded period, after which it is stopped. The ceiling is lowest on the free plan.

Hitting it usually means the task was too large rather than too slow. Split it — see Tasks, plans, sprints and lanes.

Parallel subagents

The free plan runs without parallel subagents. Paid plans allow an agent to fan work out internally, which mostly affects how quickly it explores a large codebase.

Members

Free is intended for one person evaluating the product. Professional and Enterprise do not cap members.

Adding members does not add capacity — ten people sharing one slot still get one task at a time.

Model usage

Model spend is separate from slots and is either prepaid as credits or billed directly by your own provider under BYOK. Coroid adds no markup to provider token rates.

Running out of credits stops new model calls. It does not affect your slots, projects or data.

See Model credits.

Rate limits

API and MCP access are rate limited per organization. Limits are generous for interactive use; if you are driving Coroid programmatically at volume, expect to handle 429 responses with backoff.

What happens at a limit

No limit deletes anything. Reaching one stops new work from starting, and the interface tells you which limit you reached. Existing projects, tasks, history and settings are unaffected.