Handoff context limits and recovery
Threads
A handoff leaves room for existing context, your request, attachments, instructions, tools, and later work. If the target conversation is too full, T3 Code starts a fresh provider conversation with bounded history and sends your request automatically.
T3 Code docs: portable handoffs
What it’s for
Avoid manual compaction when a long thread moves to a model with less room. Changing models or context-window options discards stale usage and compaction thresholds.
T3 Code docs: portable handoffs
When to use it
Use this when a handoff fails for lack of room or the provider rejects the input. The capacity figures are fallback estimates, not exact token counts.
T3 Code docs: portable handoffs
How to use it
If the request and attachments still leave too little room, shorten the request or select a larger-context model.
T3 Code docs: portable handoffs
Let running background work finish, or stop it, before retrying the handoff.
T3 Code docs: portable handoffs
Server operators can set T3CODE_CONTEXT_HANDOFF_TOKEN_CAP to change the initial history allowance (default 16,000; clamped to 1,024–64,000).
T3 Code docs: portable handoffs
Remember that each image reserves an estimated 8,192 tokens and other attachments reserve 4,096 each.
T3 Code docs: portable handoffs
Unofficial notes. Not affiliated with Theo or t3.gg.