Documentation
¶
Overview ¶
Package contextwindow estimates token usage for a message history and looks up the model's context-window capacity. Both numbers are approximate — exact tokenization depends on the model's tokenizer and varies enough between providers that "good enough for a watermark" is the operative bar.
Two callers consume this:
- The TUI status bar, which colors the token counter when usage crosses a configurable warn_threshold.
- The auto-summarization trigger, which fires before the next turn when usage crosses auto_threshold.
Both treat the numbers as ratios, not absolute counts, so a 10–15% estimation error is harmless.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func EstimateTokens ¶
EstimateTokens returns an approximate token count for the given messages. Uses the rough ~4-chars-per-token heuristic that holds for English text across most BPE-family tokenizers (GPT, Claude, Llama). Tool-call argument JSON and tool-result content are counted; role metadata is not (it's bounded and small).
func WindowFor ¶
WindowFor returns the context-window capacity (in tokens) for the given model tag, falling back to defaultWindow when the model isn't in the lookup table. Matching is by lowercase prefix — generous on purpose so unknown variants of known families still resolve.
The table is intentionally short. We update it as model families ship; users with an exotic or self-hosted model can always pin a fallback via context.default_window in config.toml.
Types ¶
This section is empty.
Source Files