Effort levels

Effort is a session setting with two effects:

1. a requested model reasoning level or thinking budget;
2. at ultracode, model-directed access to the bounded workflow tool.

Level	Iteron intent
low	fastest; extended thinking disabled
medium	balanced default
high	larger thinking budget
xhigh	deeper thinking budget
max	maximum built-in thinking budget
ultracode	max reasoning intent plus optional model-directed workflows

Set effort for one run:

iteron --effort high

Or use /effort in the TUI. Trusted user configuration and ITERON_EFFORT can set a default; repository configuration cannot raise or redirect effort authority.

Provider enforcement is not uniform

Adapters may apply effort exactly, map it to a smaller enum, express it through a token budget, expose only an on/off toggle, or mark it unsupported. Iteron records the requested and sent form where available instead of claiming that all providers implement the same semantics.

ultracode is not a provider model value. The provider-facing intent is max; the model may use the harness workflow tool when a task-specific graph adds value, but no workflow is launched merely because this effort level is active. See Ultracode.