opencode-acp 1.14.19-pr.314.26 → 1.14.19-pr.315.29

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAA;AA2BjD,QAAA,MAAM,MAAM,EAAE,MAwHK,CAAA;AAEnB,eAAe,MAAM,CAAA"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAA;AA2BjD,QAAA,MAAM,MAAM,EAAE,MA6IK,CAAA;AAEnB,eAAe,MAAM,CAAA"}

package/dist/index.js

@@ -2418,7 +2418,9 @@ async function saveSessionState(sessionState, logger, sessionName) {
       nextRef: sessionState.messageIds.nextRef
     },
     lastCompaction: sessionState.lastCompaction,
-    modelContextLimit: sessionState.modelContextLimit
+    modelContextLimit: sessionState.modelContextLimit,
+    modelProviderID: sessionState.modelProviderID,
+    modelID: sessionState.modelID
   };
   await writePersistedSessionState(sessionState.sessionId, state, logger);
 }
@@ -2949,6 +2951,51 @@ function applyPendingCompressionDurations(state) {
   return updates;
 }
 
+// lib/state/model-limits.ts
+function createModelLimitCatalog() {
+  const modelLimits = /* @__PURE__ */ new Map();
+  return {
+    record(providerId, modelId, limit) {
+      if (!providerId || !modelId || typeof limit !== "number" || limit <= 0) return;
+      modelLimits.set(`${providerId}/${modelId}`, limit);
+    },
+    resolve(providerId, modelId) {
+      if (!providerId || !modelId) return void 0;
+      return modelLimits.get(`${providerId}/${modelId}`);
+    },
+    /**
+     * Best-effort one-time seed from the host's provider catalog
+     * (`client.config.providers()` → GET /config/providers). Never throws;
+     * returns the number of model-limit entries recorded.
+     */
+    async hydrateFromClient(client) {
+      try {
+        const config = client;
+        const result = await config.config?.providers?.();
+        const payload = result;
+        const providers = payload?.data?.providers;
+        if (!Array.isArray(providers)) return 0;
+        let recorded = 0;
+        for (const provider of providers) {
+          const { id, models } = provider ?? {};
+          if (typeof id !== "string" || !models) continue;
+          for (const [modelId, model] of Object.entries(models)) {
+            const limit = model?.limit;
+            const context = limit?.context;
+            if (typeof context === "number" && context > 0) {
+              modelLimits.set(`${id}/${modelId}`, context);
+              recorded++;
+            }
+          }
+        }
+        return recorded;
+      } catch {
+        return 0;
+      }
+    }
+  };
+}
+
 // lib/compress/search.ts
 import { tool } from "@opencode-ai/plugin";
 async function fetchSessionMessages(client, sessionId) {
@@ -4140,56 +4187,25 @@ var SessionStateRegistry = class {
     startsByCallId: /* @__PURE__ */ new Map(),
     pendingByCallId: /* @__PURE__ */ new Map()
   };
-  // [FIX #312] Catalog of per-model context limits, keyed `${providerID}/${modelID}`.
-  // Within one LLM request the host fires experimental.chat.messages.transform
-  // BEFORE experimental.chat.system.transform (sst/opencode: session/prompt.ts
-  // triggers messages.transform, then llm/request.ts triggers system.transform
-  // during handle.process). state.modelContextLimit is written only by the
-  // system hook, so on the first request after a model switch every percentage
-  // threshold (emergencyThresholdPercent, min/maxContextLimit "%", adaptive
-  // nudge growth, GC tiers) is still computed against the PREVIOUS model's
-  // limit. This catalog lets the messages hook reconcile against the model
-  // named on the request's user message instead of waiting one turn.
-  // Entries are recorded live by the system hook every request and seeded once
-  // at plugin init from the host's /config/providers catalog.
-  modelLimits = /* @__PURE__ */ new Map();
+  // [FIX #312] Model-limit catalog (full rationale in ./model-limits.ts):
+  // lets the messages hook reconcile state.modelContextLimit against the
+  // model named on the request's user message instead of waiting one turn
+  // for the system hook. Shared implementation — the test registry stub
+  // composes the same factory.
+  catalog = createModelLimitCatalog();
   recordModelLimit(providerId, modelId, limit) {
-    if (!providerId || !modelId || typeof limit !== "number" || limit <= 0) return;
-    this.modelLimits.set(`${providerId}/${modelId}`, limit);
+    this.catalog.record(providerId, modelId, limit);
   }
   resolveModelLimit(providerId, modelId) {
-    if (!providerId || !modelId) return void 0;
-    return this.modelLimits.get(`${providerId}/${modelId}`);
+    return this.catalog.resolve(providerId, modelId);
   }
   /**
    * Best-effort one-time seed from the host's provider catalog
    * (`client.config.providers()` → GET /config/providers). Never throws;
    * returns the number of model-limit entries recorded.
    */
-  async hydrateModelLimitsFromClient(client) {
-    try {
-      const config = client;
-      const result = await config.config?.providers?.();
-      const payload = result;
-      const providers = payload?.data?.providers;
-      if (!Array.isArray(providers)) return 0;
-      let recorded = 0;
-      for (const provider of providers) {
-        const { id, models } = provider ?? {};
-        if (typeof id !== "string" || !models) continue;
-        for (const [modelId, model] of Object.entries(models)) {
-          const limit = model?.limit;
-          const context = limit?.context;
-          if (typeof context === "number" && context > 0) {
-            this.modelLimits.set(`${id}/${modelId}`, context);
-            recorded++;
-          }
-        }
-      }
-      return recorded;
-    } catch {
-      return 0;
-    }
+  hydrateModelLimitsFromClient(client) {
+    return this.catalog.hydrateFromClient(client);
   }
   get(sessionId) {
     return this.states.get(sessionId);
@@ -4279,6 +4295,8 @@ function createSessionState() {
     lastCompaction: 0,
     currentTurn: 0,
     modelContextLimit: void 0,
+    modelProviderID: void 0,
+    modelID: void 0,
     systemPromptTokens: void 0,
     qualityGateRetryPending: false
   };
@@ -4318,6 +4336,8 @@ function resetSessionState(state) {
   state.lastCompaction = 0;
   state.currentTurn = 0;
   state.modelContextLimit = void 0;
+  state.modelProviderID = void 0;
+  state.modelID = void 0;
   state.systemPromptTokens = void 0;
   state.qualityGateRetryPending = false;
 }
@@ -4392,6 +4412,8 @@ async function ensureSessionInitialized(client, state, sessionId, logger, messag
   }
   if (typeof persisted.modelContextLimit === "number" && persisted.modelContextLimit > 0) {
     state.modelContextLimit = persisted.modelContextLimit;
+    state.modelProviderID = persisted.modelProviderID;
+    state.modelID = persisted.modelID;
   }
   const applied = applyPendingCompressionDurations(state);
   if (applied > 0) {
@@ -4739,8 +4761,8 @@ ${progressBar}`;
   let toastMessage = message;
   toastMessage = config.pruneNotification === "minimal" ? toastMessage : truncateToastBody(toastMessage);
   if (config.debug) {
-    const chatMessage = config.pruneNotification === "minimal" ? message : truncateToastBody(message);
-    await sendIgnoredMessage(client, sessionId, chatMessage, params, logger);
+    logger.debug(`[ACP Debug] Compress notification:
+${message}`);
   }
   await client.tui.showToast({
     body: {
@@ -5723,11 +5745,6 @@ exact values, errors). Then add "acknowledgeRisk": true to the compress tool cal
 Without acknowledgeRisk: true, the compression will be rejected again.`;
   return new Error(message);
 }
-function buildPreemptiveAcknowledgeError() {
-  return new Error(
-    'Parameter "acknowledgeRisk": true was provided, but no quality gate rejection is pending. This parameter is only valid immediately after a compression was rejected by the quality gate. Remove it and try again.'
-  );
-}
 
 // lib/compress/pipeline.ts
 function snapshotCompressionState(state) {
@@ -6201,13 +6218,13 @@ function createCompressRangeTool(factoryCtx) {
       }
       const acknowledgeRisk = args.acknowledgeRisk === true;
       const qualityGateRetryPendingBefore = ctx.state.qualityGateRetryPending;
-      if (acknowledgeRisk && !ctx.state.qualityGateRetryPending) {
-        throw buildPreemptiveAcknowledgeError();
+      const bypassQuality = acknowledgeRisk && ctx.state.qualityGateRetryPending;
+      const ignoredAcknowledgeRisk = acknowledgeRisk && !ctx.state.qualityGateRetryPending;
+      if (ignoredAcknowledgeRisk) {
+        ctx.logger.warn("compress: acknowledgeRisk ignored \u2014 no quality gate rejection pending");
       }
-      if (acknowledgeRisk) {
-        ctx.state.qualityGateRetryPending = false;
-      } else {
-        ctx.state.qualityGateRetryPending = false;
+      ctx.state.qualityGateRetryPending = false;
+      if (!bypassQuality) {
         for (const plan of preparedPlans) {
           const result = evaluatePreCommitQuality(
             rawMessages,
@@ -6283,7 +6300,10 @@ function createCompressRangeTool(factoryCtx) {
       const skippedNote = phantomSkipNotice !== null ? `
 \u26A0\uFE0F ${phantomSkipNotice}
 ` : "";
-      return `Compressed ${totalCompressedMessages} messages into ${COMPRESSED_BLOCK_HEADER}.${skippedNote}
+      const ackNote = ignoredAcknowledgeRisk ? `
+\u26A0\uFE0F acknowledgeRisk was ignored: no quality gate rejection was pending, so quality checks ran normally. Only pass it when retrying immediately after a quality gate rejection.
+` : "";
+      return `Compressed ${totalCompressedMessages} messages into ${COMPRESSED_BLOCK_HEADER}.${skippedNote}${ackNote}
 IMPORTANT: This was an automatic context compression. You MUST continue your previous task exactly where you left off. Do NOT ask the user what to do next.
 \u{1F4A1} Tip: Use search_context('keyword') to find compressed content when you need it later.`;
     }
@@ -7617,16 +7637,6 @@ var injectCompressNudges = (state, config, logger, messages, prompts, compressio
     { logger }
   );
   const hasRecommendations = recommendedRanges.length > 0;
-  if (config.debug && contextRanges.compressible.length > 0) {
-    const compressible = contextRanges.compressible;
-    const fmt = (n) => n >= 1e3 ? `${(n / 1e3).toFixed(1)}K` : String(n);
-    const lines = [
-      `[ACP Debug] Recommendation filter:`,
-      `  Input: ${compressible.length} range(s), ${fmt(compressible.reduce((s, r) => s + r.tokens, 0))} tokens`,
-      `  Output: ${recommendedRanges.length} range(s) (last segment marked dangerous)`
-    ];
-    logger.debug(lines.join("\n"));
-  }
   const allProtected = contextRanges.compressible.length === 0 && contextRanges.protected.length > 0;
   const allInProtectedZone = protectedRefs.size > 0 && unprotectedCompressible.length === 0;
   const nothingToCompress = allProtected || allInProtectedZone;
@@ -7722,6 +7732,16 @@ ${rules}`;
     }
   }
   state.nudges.shouldInjectThisTurn = shouldInject;
+  if (shouldInject && config.debug && contextRanges.compressible.length > 0) {
+    const compressible = contextRanges.compressible;
+    const fmt = (n) => n >= 1e3 ? `${(n / 1e3).toFixed(1)}K` : String(n);
+    const lines = [
+      `[ACP Debug] Recommendation filter:`,
+      `  Input: ${compressible.length} range(s), ${fmt(compressible.reduce((s, r) => s + r.tokens, 0))} tokens`,
+      `  Output: ${recommendedRanges.length} range(s) (last segment marked dangerous)`
+    ];
+    logger.debug(lines.join("\n"));
+  }
   let tipsText = null;
   if (shouldInject) {
     if (suffixMessage && composition.total > 0) {
@@ -8176,10 +8196,6 @@ function resolveSingleBlockTarget(messagesState, blockIdArg) {
         error: `Error: Block ${target.displayId} is nested inside active block ${activeAncestorBlockId}. Decompress block ${activeAncestorBlockId} first.`
       };
     }
-    return {
-      ok: false,
-      error: `Error: Block ${target.displayId} is not active. It may have already been decompressed.`
-    };
   }
   return { ok: true, targets: [target] };
 }
@@ -8332,7 +8348,7 @@ function createDecompressTool(factoryCtx) {
         const blockMessages = rawMessages.filter((m) => msgIdSet.has(extractMessageId(m)));
         const lines2 = blockMessages.map(extractMessageText2);
         const { writeFile: writeFile3 } = await import("fs/promises");
-        const fileContent = lines2.length > 0 ? lines2.join("\n\n---\n\n") : activeBlocks[0]?.summary ?? "(no content available)";
+        const fileContent = lines2.length > 0 ? lines2.join("\n\n---\n\n") : targets[0]?.blocks[0]?.summary ?? "(no content available)";
         await writeFile3(targetPath, fileContent, "utf-8");
         const displayIds2 = targets.map((t) => `b${t.displayId}`).join(", ");
         return `Block(s) ${displayIds2} content (${blockMessages.length} messages, ${fileContent.length} chars) written to ${targetPath}. Block(s) stay compressed \u2014 context unchanged. Use read tool to access specific parts.`;
@@ -8789,16 +8805,23 @@ function renderCompressedDrilldown(blocks, sort, limit, blocksById) {
   lines.push(`Sorted by ${sort === "time" ? "time" : sort === "age" ? "age" : "size"}`);
   lines.push("");
   const shown = sorted.slice(0, limit);
+  const activeCount = sorted.filter((b) => b.active).length;
+  const inactiveCount = sorted.length - activeCount;
+  if (inactiveCount > 0) {
+    lines.push(`${activeCount} active, ${inactiveCount} inactive/consumed`);
+    lines.push("");
+  }
   for (const b of shown) {
     const survived = b.survivedCount ?? 0;
     const gen = b.generation ?? "young";
     const effCount = b.effectiveMessageIds?.length ?? 0;
     const consumed = b.includedBlockIds && b.includedBlockIds.length > 0 ? ` nested=[${b.includedBlockIds.map((n) => `b${n}`).join(",")}]` : "";
+    const status = b.active ? "" : " [inactive]";
     const topic = b.topic || "(no topic)";
     const tier = tierLabel(b);
     const effTokens = getEffectiveCompressedTokens(b, blocksById);
     lines.push(
-      `  b${b.blockId} (${tier})  ${formatTokens(effTokens)}\u2192${formatTokens(b.summaryTokens)}  ${formatAge(b.createdAt)}  ${formatIdRange(b)}  age=${survived} ${gen} eff=${effCount}${consumed}`
+      `  b${b.blockId} (${tier})  ${formatTokens(effTokens)}\u2192${formatTokens(b.summaryTokens)}  ${formatAge(b.createdAt)}  ${formatIdRange(b)}  age=${survived} ${gen} eff=${effCount}${consumed}${status}`
     );
     lines.push(`    "${topic}"`);
   }
@@ -8819,8 +8842,10 @@ function buildStatusReport(renderCtx, rawMessages, options) {
   const sort = options?.sort ?? "size";
   const limit = options?.limit ?? 30;
   const msgState = renderCtx.state.prune.messages;
-  const activeIds = Array.from(msgState.activeBlockIds).sort((a, b) => a - b);
-  const allBlocks = activeIds.map((id) => msgState.blocksById.get(id)).filter((b) => b !== void 0 && b.active);
+  const allBlocks = Array.from(msgState.blocksById.values()).sort(
+    (a, b) => a.blockId - b.blockId
+  );
+  const activeBlocks = allBlocks.filter((b) => b.active);
   const lines = [];
   if (scope === "compressed") {
     lines.push(...renderCompressedDrilldown(allBlocks, sort, limit, msgState.blocksById));
@@ -8842,7 +8867,7 @@ function buildStatusReport(renderCtx, rawMessages, options) {
         visibleMsgs,
         summaryTokens,
         systemTokens,
-        allBlocks,
+        activeBlocks,
         false,
         rawMessages,
         renderCtx
@@ -9000,7 +9025,7 @@ import { writeFile as writeFile2, mkdir as mkdir2 } from "fs/promises";
 import { join as join3 } from "path";
 import { existsSync as existsSync3 } from "fs";
 import { homedir as homedir3 } from "os";
-var LOG_VERSION = true ? "1.14.19-pr.314.26" : "dev";
+var LOG_VERSION = true ? "1.14.19-pr.315.29" : "dev";
 var Logger = class {
   logDir;
   enabled;
@@ -10575,6 +10600,8 @@ function createSystemPromptHandler(registry4, logger, config, prompts) {
     const state = input.sessionID ? registry4.get(input.sessionID) : void 0;
     if (state && input.model?.limit?.context) {
       state.modelContextLimit = input.model.limit.context;
+      state.modelProviderID = input.model?.providerID;
+      state.modelID = input.model?.id;
     }
     if (!state || state.isSubAgent && !config.allowSubAgents) {
       return;
@@ -10634,6 +10661,22 @@ function createChatMessageTransformHandler(client, registry4, logger, config, pr
       );
       if (requestModelLimit !== void 0) {
         state.modelContextLimit = requestModelLimit;
+        state.modelProviderID = requestModel?.providerID;
+        state.modelID = requestModel?.modelID;
+      } else if (
+        // [FIX #312 fallback] Catalog miss: we cannot CORRECT the
+        // limit, but we can tell when it belongs to a DIFFERENT model.
+        // Invalidate instead of letting every percentage threshold
+        // below run against the wrong window (#312's false positive).
+        // States persisted before this identity pair existed carry no
+        // identity and are treated as stale for the same reason.
+        // Consumers already tolerate undefined — fresh sessions run
+        // with it until the first system.transform sets the pair.
+        requestModel?.providerID && requestModel?.modelID && state.modelContextLimit !== void 0 && (state.modelProviderID !== requestModel.providerID || state.modelID !== requestModel.modelID)
+      ) {
+        state.modelContextLimit = void 0;
+        state.modelProviderID = requestModel.providerID;
+        state.modelID = requestModel.modelID;
       }
       await updatePerTurnState(state, logger, messages);
     }
@@ -10680,23 +10723,6 @@ function createChatMessageTransformHandler(client, registry4, logger, config, pr
       config.debug ? (text) => {
         logger.debug(`[ACP Debug] Nudge injected:
 ${text}`);
-        if (state.sessionId && lastUserMessage) {
-          const userInfo = lastUserMessage.info;
-          sendIgnoredMessage(
-            client,
-            state.sessionId,
-            `[ACP Debug Nudge]
-${text}`,
-            {
-              providerId: userInfo.model?.providerID,
-              modelId: userInfo.model?.modelID,
-              agent: userInfo.agent,
-              variant: userInfo.variant
-            },
-            logger
-          ).catch(() => {
-          });
-        }
         client.tui.showToast({
           body: {
             title: "ACP: Nudge Injected",
@@ -11020,8 +11046,25 @@ var server = (async (ctx) => {
   if (isSecureMode()) {
     configureClientAuth(ctx.client);
   }
-  registry4.hydrateModelLimitsFromClient(ctx.client).catch(() => {
-  });
+  registry4.hydrateModelLimitsFromClient(ctx.client).then(
+    (recorded) => {
+      if (recorded > 0) {
+        logger.info("Model limit catalog seeded from provider config", {
+          models: recorded
+        });
+      } else {
+        logger.warn(
+          "Model limit catalog seeding recorded no entries \u2014 falling back to per-request refresh (system.transform)"
+        );
+      }
+    },
+    (error) => {
+      logger.warn(
+        "Model limit catalog seeding failed \u2014 falling back to per-request refresh (system.transform)",
+        { error: error instanceof Error ? error.message : String(error) }
+      );
+    }
+  );
   logger.info("DCP initialized");
   startAutoUpdate(ctx, config.autoUpdate);
   const compressToolContext = {