Context recovery (1.9.1, 1.13.0) shrinks what a turn sends, then retries. It assumes the reason the provider said "too big" is history size. That is usually true, and when it is not, recovery loops.
Two non-history causes, both real, both now detected before any reduction runs:
Output capacity. A large max_tokens reservation can push the request over the window. Cutting
history does not help, because the reservation is the problem, so Pengy lowers the cap on
max_output_tokens instead of burning four reduction passes on a request that cannot fit.
The newest tool result. A result from the turn the model is about to act on is the payload. It must not be cut, so it is pinned: the newest result is never reduced, and the ladder trims older ones instead.
The 1.11.1 output-truncation fix was extended at the same time: a stream ending with
finish_reason: length now fails in every edition (the Rust and C++ editions had it in the CLI
library; the GUIs and web UIs did not, and C++ was silent in the GUI and web surfaces), the
max_output_tokens config is honoured in the Rust engine, and the notice points at the cap as the
thing to raise.
There is a small honesty fix in here I like: the budget used to be estimated from the conversation text alone. The tool schema is about 18.5k tokens, so the estimator was blind to the largest single contributor to the request. It counts tool definitions now.