billion-context-omp 0.2.3 → 0.2.4

package/dist/auto-compress.d.ts

@@ -43,8 +43,11 @@ export declare function buildSummaryPrompt(prompts: Prompts): string;
  *  deliberately NOT used here: fold blocks only replay from in-stream
  *  compress tool calls, which the truncation removes. `previousSummary` (an
  *  earlier compaction's summary) is folded in so iterative compactions never
- *  drop it. Returns null on any failure so the caller falls back to Pi's
- *  native compaction. */
+ *  drop it. On a failed attempt the LLM call is
+ *  retried ONCE (stochastic formatting failures recover on a fresh call);
+ *  returns null only when the retry also fails — the caller then CANCELS the
+ *  compaction (no native fallback: ACP owns compression, host-default
+ *  summaries are the failure mode this hook exists to prevent). */
 export declare function summarizeMessages(ctx: ExtensionContext, messages: AgentMessage[], prompts: Prompts, configuredModel?: string | null, opts?: {
     previousSummary?: string;
     customInstructions?: string;
@@ -58,11 +61,9 @@ export declare function summarizeMessages(ctx: ExtensionContext, messages: Agent
     summary: string;
     model: string;
 } | null>;
-/** Generate a summary for a message range using the compression model. Shared
- *  entry point for the `/compact` handler. Returns null when no model is
- *  usable, the slice is empty, the model is unauthenticated, or the response
- *  is unparseable — the caller then returns `undefined` so Pi falls back to
- *  its native compaction. */
+/** Generate a summary for a message range using the compression model.
+ *  Currently unused by the extension (kept as the shared range-summary
+ *  surface); null = hard failure, caller decides. */
 export declare function summarizeRange(ctx: ExtensionContext, messages: CoreMessage[], state: CompressionState, startRef: string, endRef: string, prompts: Prompts, configuredModel?: string | null): Promise<{
     summary: string;
     model: string;

package/dist/index.js

@@ -4558,7 +4558,7 @@ async function statusReport(runtime, ctx) {
   const coveredIds = collectCoveredMessageIds(state);
   const sentTokens = estimateTokens(coreMessages, coveredIds) + systemPromptTokens;
   const turn = runtime.core.processTurn({ messages: coreMessages, state, config, tokenCount: sentTokens });
-  const versionStr = "0.2.3" ? `billion-context-omp@${"0.2.3"}` : void 0;
+  const versionStr = "0.2.4" ? `billion-context-omp@${"0.2.4"}` : void 0;
   return buildStatusPanel({
     version: versionStr,
     tokenCount: sessionTokens,
@@ -4807,7 +4807,7 @@ import { join as join3 } from "path";
 import { complete } from "@oh-my-pi/pi-ai";
 import { CONFIG_DIR_NAME as CONFIG_DIR_NAME2 } from "@oh-my-pi/pi-utils";
 var TIMEOUT_MS = 6e4;
-var MAX_OUTPUT_TOKENS = 3e3;
+var MAX_OUTPUT_TOKENS = 8e3;
 var MAX_SLICE_CHARS = 15e4;
 var MAX_MSG_CHARS = 4e3;
 function readCompressModel() {
@@ -4857,9 +4857,10 @@ function parseSummary(text) {
   try {
     const obj = JSON.parse(cleaned);
     if (typeof obj.summary === "string" && obj.summary.length > 0) return obj.summary;
+    return null;
   } catch {
   }
-  return null;
+  return cleaned.length >= 50 ? cleaned : null;
 }
 function buildSummaryPrompt(prompts) {
   return prompts.compressPhilosophy.trim() + "\n\n" + prompts.howToCompressRules.trim() + '\n\nCompress the message range provided below into ONE dense, self-contained technical summary following the rules above. Output ONLY a JSON object: {"summary": "..."} where the value is the full summary as a single string.';
@@ -4896,14 +4897,32 @@ User instructions for this compaction: ${custom}`;
     const userText = `ENTIRE conversation to compress (${slice.length} messages, ~${tokens} tokens). Compress it:
 
 ` + formatSlice(slice, opts?.messageRefs ? { ...createInitialState(), messageRefs: opts.messageRefs } : createInitialState());
-    const response = await run(
-      model,
-      { systemPrompt: [instructions], messages: [{ role: "user", content: [{ type: "text", text: userText }], timestamp: Date.now() }] },
-      { apiKey: auth.apiKey, headers: auth.headers, maxTokens: MAX_OUTPUT_TOKENS, signal: ac.signal }
-    );
-    const summary = parseSummary(
-      response.content.filter((c) => c.type === "text").map((c) => c.text).join("\n")
-    );
+    const attempt = async () => {
+      const response = await run(
+        model,
+        { systemPrompt: [instructions], messages: [{ role: "user", content: [{ type: "text", text: userText }], timestamp: Date.now() }] },
+        { apiKey: auth.apiKey, headers: auth.headers, maxTokens: MAX_OUTPUT_TOKENS, signal: ac.signal }
+      );
+      return parseSummary(
+        response.content.filter((c) => c.type === "text").map((c) => c.text).join("\n")
+      );
+    };
+    let summary = null;
+    try {
+      summary = await attempt();
+    } catch (e) {
+      if (opts?.signal?.aborted || ac.signal.aborted) throw e;
+      logWarn("summarize-messages", { event: "attempt-failed", model: label, error: String(e) });
+    }
+    if (!summary) {
+      try {
+        summary = await attempt();
+        if (summary) logInfo("summarize-messages", { event: "recovered-on-retry", model: label, messages: slice.length });
+      } catch (e) {
+        if (opts?.signal?.aborted || ac.signal.aborted) throw e;
+        logWarn("summarize-messages", { event: "retry-failed", model: label, error: String(e) });
+      }
+    }
     if (!summary) {
       logWarn("summarize-messages", { event: "unparseable-summary", model: label, messages: slice.length });
       return null;
@@ -5266,7 +5285,7 @@ async function checkForUpdate(autoUpdate, notify) {
     const data = await res.json();
     const latest = data.version;
     if (!latest) return;
-    const current = runtimeVersion ?? "0.2.3";
+    const current = runtimeVersion ?? "0.2.4";
     const hasUpdate = isNewer(latest, current);
     debug.event("update-check", {
       current,
@@ -5555,8 +5574,8 @@ function wireCompactionDisable(pi, runtime) {
         messageRefs: slot.state.messageRefs
       });
       if (!result) {
-        ctx.ui?.notify?.("ACP: compression fell back to Pi native compaction", "warning");
-        return void 0;
+        ctx.ui?.notify?.("ACP: /compact aborted \u2014 summary generation failed after retry (details in ~/.omp/acp-omp.log)", "error");
+        return { cancel: true };
       }
       logInfo("compact", {
         sid,
@@ -5583,9 +5602,9 @@ function wireCompactionDisable(pi, runtime) {
 function wireSessionLifecycle(pi, runtime) {
   pi.on("session_start", async (_event, ctx) => {
     const sid = ctx.sessionManager.getSessionId();
-    logInfo("session", { event: "start", sid, cwd: ctx.cwd, debug: runtime.adapter.debug ?? null, version: true ? "0.2.3" : null });
+    logInfo("session", { event: "start", sid, cwd: ctx.cwd, debug: runtime.adapter.debug ?? null, version: true ? "0.2.4" : null });
     const selfPath = import.meta.url;
-    const conflict = stampAndDetect(selfPath, true ? "0.2.3" : null);
+    const conflict = stampAndDetect(selfPath, true ? "0.2.4" : null);
     if (conflict) {
       logWarn("instance", { event: "dual-instance", self: selfPath, other: conflict.path, otherPid: conflict.pid, otherVersion: conflict.version });
       try {