opencode-acp 1.5.0 → 1.5.1

package/dist/index.js

@@ -894,6 +894,7 @@ var VALID_CONFIG_KEYS = /* @__PURE__ */ new Set([
   "compress.modelMaxLimits",
   "compress.modelMinLimits",
   "compress.nudgeFrequency",
+  "compress.perMessageNudgeGrowthPercent",
   "compress.iterationNudgeThreshold",
   "compress.nudgeForce",
   "compress.protectedTools",
@@ -1117,6 +1118,13 @@ function validateConfigTypes(config) {
           actual: `${compress.nudgeFrequency} (will be clamped to 1)`
         });
       }
+      if (compress.perMessageNudgeGrowthPercent !== void 0 && typeof compress.perMessageNudgeGrowthPercent !== "number") {
+        errors.push({
+          key: "compress.perMessageNudgeGrowthPercent",
+          expected: "number",
+          actual: typeof compress.perMessageNudgeGrowthPercent
+        });
+      }
       if (compress.iterationNudgeThreshold !== void 0 && typeof compress.iterationNudgeThreshold !== "number") {
         errors.push({
           key: "compress.iterationNudgeThreshold",
@@ -1461,6 +1469,7 @@ var defaultConfig = {
     maxContextLimit: "55%",
     minContextLimit: "45%",
     nudgeFrequency: 5,
+    perMessageNudgeGrowthPercent: 3,
     iterationNudgeThreshold: 15,
     nudgeForce: "soft",
     protectedTools: [...COMPRESS_DEFAULT_PROTECTED_TOOLS],
@@ -1612,6 +1621,7 @@ function mergeCompress(base, override) {
     modelMaxLimits: override.modelMaxLimits ?? base.modelMaxLimits,
     modelMinLimits: override.modelMinLimits ?? base.modelMinLimits,
     nudgeFrequency: override.nudgeFrequency ?? base.nudgeFrequency,
+    perMessageNudgeGrowthPercent: override.perMessageNudgeGrowthPercent ?? base.perMessageNudgeGrowthPercent,
     iterationNudgeThreshold: override.iterationNudgeThreshold ?? base.iterationNudgeThreshold,
     nudgeForce: override.nudgeForce ?? base.nudgeForce,
     protectedTools: [.../* @__PURE__ */ new Set([...base.protectedTools, ...override.protectedTools ?? []])],
@@ -3019,7 +3029,9 @@ function resetOnCompaction(state) {
   state.nudges = {
     contextLimitAnchors: /* @__PURE__ */ new Set(),
     turnNudgeAnchors: /* @__PURE__ */ new Set(),
-    iterationNudgeAnchors: /* @__PURE__ */ new Set()
+    iterationNudgeAnchors: /* @__PURE__ */ new Set(),
+    lastPerMessageNudgeTurn: 0,
+    lastPerMessageNudgeTokens: 0
   };
   state.messageIds = {
     byRawId: /* @__PURE__ */ new Map(),
@@ -3085,7 +3097,9 @@ async function saveSessionState(sessionState, logger, sessionName) {
     nudges: {
       contextLimitAnchors: Array.from(sessionState.nudges.contextLimitAnchors),
       turnNudgeAnchors: Array.from(sessionState.nudges.turnNudgeAnchors),
-      iterationNudgeAnchors: Array.from(sessionState.nudges.iterationNudgeAnchors)
+      iterationNudgeAnchors: Array.from(sessionState.nudges.iterationNudgeAnchors),
+      lastPerMessageNudgeTurn: sessionState.nudges.lastPerMessageNudgeTurn ?? 0,
+      lastPerMessageNudgeTokens: sessionState.nudges.lastPerMessageNudgeTokens ?? 0
     },
     stats: sessionState.stats,
     lastUpdated: (/* @__PURE__ */ new Date()).toISOString(),
@@ -3299,7 +3313,9 @@ function createSessionState() {
     nudges: {
       contextLimitAnchors: /* @__PURE__ */ new Set(),
       turnNudgeAnchors: /* @__PURE__ */ new Set(),
-      iterationNudgeAnchors: /* @__PURE__ */ new Set()
+      iterationNudgeAnchors: /* @__PURE__ */ new Set(),
+      lastPerMessageNudgeTurn: 0,
+      lastPerMessageNudgeTokens: 0
     },
     stats: {
       pruneTokenCounter: 0,
@@ -3336,7 +3352,9 @@ function resetSessionState(state) {
   state.nudges = {
     contextLimitAnchors: /* @__PURE__ */ new Set(),
     turnNudgeAnchors: /* @__PURE__ */ new Set(),
-    iterationNudgeAnchors: /* @__PURE__ */ new Set()
+    iterationNudgeAnchors: /* @__PURE__ */ new Set(),
+    lastPerMessageNudgeTurn: 0,
+    lastPerMessageNudgeTokens: 0
   };
   state.stats = {
     pruneTokenCounter: 0,
@@ -3381,6 +3399,8 @@ async function ensureSessionInitialized(client, state, sessionId, logger, messag
   state.nudges.iterationNudgeAnchors = new Set(
     persisted.nudges.iterationNudgeAnchors || []
   );
+  state.nudges.lastPerMessageNudgeTurn = persisted.nudges.lastPerMessageNudgeTurn ?? 0;
+  state.nudges.lastPerMessageNudgeTokens = persisted.nudges.lastPerMessageNudgeTokens ?? 0;
   state.stats = {
     pruneTokenCounter: persisted.stats?.pruneTokenCounter || 0,
     totalPruneTokens: persisted.stats?.totalPruneTokens || 0
@@ -4815,7 +4835,8 @@ var createSyntheticUserMessage = (baseMessage, content, stableSeed) => {
         sessionID: userInfo.sessionID,
         messageID: messageId,
         type: "text",
-        text: content
+        text: content,
+        synthetic: true
       }
     ]
   };
@@ -5225,12 +5246,18 @@ function buildCompressedBlockGuidance(state, gcConfig, context) {
     const recent = refs.slice(-20).join(", ");
     blockList = `${recent} (+${blockCount - 20} older, use decompress to access by ID)`;
   }
+  const includeHint = context?.includeHint ?? true;
   const lines = [
     "Compressed block context:",
     `- Active compressed blocks: ${blockCount} (${blockList})`,
-    "- If your selected compression range includes any listed block, include each required placeholder exactly once in the summary using `(bN)`.",
-    "- \u{1F4A1} When you've finished using tool outputs, compress them \u2014 you can decompress later if needed. Lean context improves accuracy."
+    "- If your selected compression range includes any listed block, include each required placeholder exactly once in the summary using `(bN)`."
   ];
+  if (includeHint) {
+    lines.push("- \u{1F4A1} When you've finished using tool outputs, compress them \u2014 you can decompress later if needed. Lean context improves accuracy.");
+  }
+  if (blockCount > 50) {
+    lines.push(`- \u{1F500} You have ${blockCount} blocks \u2014 consider merging adjacent same-topic blocks instead of finding new content to compress. This permanently reduces per-turn overhead.`);
+  }
   const usageRatio = context?.currentTokens && context?.modelContextLimit ? context.currentTokens / context.modelContextLimit : 0;
   if (gcConfig && usageRatio > 0.5) {
     const promotionThreshold = gcConfig.promotionThreshold;
@@ -5573,7 +5600,7 @@ function resolveThresholdPercent(threshold, modelContextLimit) {
   const parsed = parseFloat(threshold);
   return isNaN(parsed) ? void 0 : parsed;
 }
-function buildContextUsageGuidance(config, currentTokens, modelContextLimit) {
+function buildContextUsageGuidance(config, currentTokens, modelContextLimit, minimal = false) {
   if (currentTokens === void 0 || modelContextLimit === void 0 || modelContextLimit === 0) {
     return "";
   }
@@ -5583,9 +5610,14 @@ function buildContextUsageGuidance(config, currentTokens, modelContextLimit) {
   const minPct = resolveThresholdPercent(config.compress.minContextLimit, modelContextLimit) ?? 45;
   const maxPct = resolveThresholdPercent(config.compress.maxContextLimit, modelContextLimit) ?? 55;
   const base = `Context usage: ${formatK(currentTokens)} / ${formatK(modelContextLimit)} tokens (${percentage}%).`;
+  if (minimal) {
+    return `
+
+${base}`;
+  }
   let guidance;
   if (pct < minPct) {
-    guidance = " \u{1F4A1} Be frugal with context \u2014 compress tool outputs you've finished using into summaries. You can decompress later; nothing is permanently lost. Lean context means better accuracy. Extract and keep what matters: user intent, key decisions, file paths, and important findings \u2014 even if buried in large messages. Compress everything else, including verbose parts of any message.";
+    guidance = " \u{1F4A1} Be frugal with context \u2014 if you see large completed outputs (>2000 tokens), compress them into summaries. If everything is already compressed, skip this nudge. You can decompress later if needed. Extract and keep what matters: user intent, key decisions, file paths, and important findings. Compress everything else.";
   } else if (pct < maxPct) {
     guidance = " \u26A0\uFE0F Context is growing \u2014 compress completed sections and high-token waste now. Preserve key details.";
   } else {
@@ -5687,6 +5719,18 @@ function createSuffixMessage(messages) {
   messages.push(synthetic);
   return synthetic;
 }
+function shouldInjectPerMessageNudge(state, config, currentTokens, modelContextLimit) {
+  const turn = state.currentTurn ?? 0;
+  const lastTurn = state.nudges.lastPerMessageNudgeTurn ?? 0;
+  const turnsSinceLast = turn - lastTurn;
+  const tokens = currentTokens ?? 0;
+  const lastTokens = state.nudges.lastPerMessageNudgeTokens ?? 0;
+  const tokenGrowth = tokens - lastTokens;
+  const tokenGrowthPercent = modelContextLimit ? tokenGrowth / modelContextLimit * 100 : 0;
+  const frequency = config.compress.nudgeFrequency ?? 5;
+  const growthThreshold = config.compress.perMessageNudgeGrowthPercent ?? 3;
+  return turnsSinceLast >= frequency || tokenGrowthPercent >= growthThreshold;
+}
 var injectCompressNudges = (state, config, logger, messages, prompts, compressionPriorities) => {
   if (compressPermission(state, config) === "deny") {
     return;
@@ -5771,21 +5815,26 @@ var injectCompressNudges = (state, config, logger, messages, prompts, compressio
   }
   const suffixMessage = createSuffixMessage(messages);
   applyAnchoredNudges(state, config, messages, prompts, compressionPriorities, currentTokens, modelContextLimit, suffixMessage);
-  injectContextUsage(suffixMessage, config, currentTokens, modelContextLimit);
+  const shouldNudge = shouldInjectPerMessageNudge(state, config, currentTokens, modelContextLimit);
+  injectContextUsage(suffixMessage, config, currentTokens, modelContextLimit, !shouldNudge);
   if (config.compress.mode !== "message") {
-    const blockGuidance = buildCompressedBlockGuidance(state, config.gc, { currentTokens, modelContextLimit });
+    const blockGuidance = buildCompressedBlockGuidance(state, config.gc, { currentTokens, modelContextLimit, includeHint: shouldNudge });
     if (blockGuidance.trim() && suffixMessage) {
       appendToLastTextPart(suffixMessage, "\n\n" + blockGuidance);
     }
   }
+  if (shouldNudge) {
+    state.nudges.lastPerMessageNudgeTurn = state.currentTurn ?? 0;
+    state.nudges.lastPerMessageNudgeTokens = currentTokens ?? 0;
+  }
   injectVisibleIdRange(state, messages, suffixMessage);
   if (anchorsChanged) {
     void saveSessionState(state, logger);
   }
 };
-function injectContextUsage(target, config, currentTokens, modelContextLimit) {
+function injectContextUsage(target, config, currentTokens, modelContextLimit, minimal = false) {
   if (!target) return;
-  const usageTag = buildContextUsageGuidance(config, currentTokens, modelContextLimit);
+  const usageTag = buildContextUsageGuidance(config, currentTokens, modelContextLimit, minimal);
   if (!usageTag) return;
   for (const part of target.parts) {
     if (part.type === "text") {