上下文压缩

def _trigger_token_threshold(self, context: ModelContext) -> int:
    budget = effective_context_budget(context, model_config=self._round_config.model)
    return max(int(budget * self._trigger_context_ratio), 1)

# —— 调用处（同文件 222-228 行）——
trigger_threshold = self._trigger_token_threshold(context)
if total_tokens >= trigger_threshold:
    logger.info(
        f"[{self.processor_type()} triggered] estimated context window tokens {total_tokens} "
        f"reaches trigger_context_ratio {self._trigger_context_ratio} threshold {trigger_threshold}"
    )
    return True

/// Runs token-budget inline auto-compaction as a normal compaction lifecycle.
///
/// Token-budget compaction skips model/server summarization and installs a fresh context window
/// instead. It is still modeled as compaction so compact hooks and `ContextCompaction` turn items
/// observe the same lifecycle as local or remote compaction.
pub(crate) async fn run_inline_auto_compact_task(

// —— 配置层（core/src/config/mod.rs 621-627 行）——
/// Token usage threshold triggering auto-compaction of conversation history.
pub model_auto_compact_token_limit: Option<i64>,
/// Controls whether `model_auto_compact_token_limit` applies to the full
/// active context or only tokens after the carried compaction-window prefix.
pub model_auto_compact_token_limit_scope: AutoCompactTokenLimitScope,

const modelPolicy: z<ModelCompactPolicyConfig> = z.object({
  provider: z.string().required(),
  model: z.string().required(),
  thresholdRatio: thresholdRatioSchema,
  retainRatio: retainRatioSchema,
  retainTokens: retainTokensSchema,
  summarizationProvider: summarizationProviderSchema,
  summarizationModel: summarizationModelSchema,
  maxTokens: maxTokensSchema,
  compactionRetries: compactionRetriesSchema,
  maxOverflowRetries: maxOverflowRetriesSchema,

/** Return whether context usage exceeds the configured compaction threshold. */
export function shouldCompact(contextTokens: number, contextWindow: number, settings: CompactionSettings): boolean {
	if (!settings.enabled) return false;
	return contextTokens > contextWindow - settings.reserveTokens;
}

/** Default compaction settings used by the harness.（同文件 157-162 行）*/
export const DEFAULT_COMPACTION_SETTINGS: CompactionSettings = {
	enabled: true,
	reserveTokens: 16384,
	keepRecentTokens: 20000,
};