pi-tuck 1.0.11 → 1.0.12

package/CHANGELOG.md

@@ -3,6 +3,33 @@
 All notable changes to pi-tuck. The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and versions follow
 [Semantic Versioning](https://semver.org/).
 
+## [1.0.12] - 2026-11-06
+
+### Fixed
+
+- **A prompt could be lost, with no notice (1.0.10 and 1.0.11).** When a new prompt arrived over the cut point and pi-tuck's cut kept nothing of the
+  conversation before it, which happens when everything since the last cut is one long turn, the prompt was dropped from the request and from the
+  saved session. The model never saw it and usually went on with the previous task, so work you asked for could be lost, and could look answered.
+  The next prompt ran normally. A fresh one-prompt `pi -p` run could not hit it; a run that resumed a long session could. If you used 1.0.10 or 1.0.11
+  and a reply ignored what you had just asked, this is the likely cause. Now the prompt always reaches the model and stays in the session.
+
+### Added
+
+- The settings stress test now fails any prompt missing from a request it opened, or from the session after it: 13 such findings on 1.0.11, none here.
+
+### Known issues
+
+- Correction to the 1.0.11 notes: their count of 4 failing stress cells of 272 understated cut rhythm. With setups that leave a real summary and tail
+  after a cut, 1.0.11 cuts too often at small windows, freeing little each time: about 99 cuts in 200 tool rounds at an 81,920 window with a 24k setup.
+  It costs time and extra work. An automatic cut point that fixes it is being tested and is not in this release.
+- The known issues listed under 1.0.11 still apply.
+
+### Release gates
+
+- **Shipped without a live gate, as a defect that loses work.** Tests and CI passed, and the stress test passed on Pi 1.2.0 and the newest Pi.
+- Replay differs from 1.0.11, as it has to: 1,124 of 13,946 requests on 29 recordings, each change starting at the first request of a run, with 0 dead
+  tasks, 0 refused requests and the same 9 clips.
+
 ## [1.0.11] - 2026-11-06
 
 ### Fixed
@@ -12,7 +39,7 @@ All notable changes to pi-tuck. The format follows [Keep a Changelog](https://ke
 - A fresh tool result is clipped in more of the cases where the request would otherwise be refused. The clip's budget is now counted in the
   same units as the request and allows for what the smallest cut keeps, so a round that fits only when clipped no longer ends the run. This
   is most of the refusals the settings stress test found.
-- One wrong usage report no longer refuses later prompts: a reply that reports more than the model's window is not learned from.
+- A reply that reports more than the model's window no longer raises pi-tuck's estimate for later requests.
 - When a provider refuses a request as over its window, pi-tuck now says so in a notice, where the turn could read as a hang.
 - Every refusal says why.
 
@@ -30,8 +57,8 @@ All notable changes to pi-tuck. The format follows [Keep a Changelog](https://ke
   - With a setup near the band, cuts can come too often: with Pi's defaults, a 40k setup at a 65,536 window made 25 cuts in 60 tool rounds, each
     freeing little. It costs extra work, not a task. One sampled cell at a 167,739 window also trips this count. The fix is planned for the next
     patch.
-  - A wrong usage report at or under the window can still raise the estimate for a while: in the stress cell for it, a 6,000-token prompt is
-    refused in a 28,672-token room.
+  - A reply that reports more than the window reads to Pi as a silent overflow, so Pi asks for a recovery cut. When nothing can be cut, pi-tuck
+    refuses the next prompt: in the stress cell for it, a 6,000-token prompt in a 28,672-token room. The prompt after that runs.
 - If a provider keeps rejecting requests in a wording Pi does not class as a context overflow, or with a bare HTTP 413, pi-tuck does not recover
   on its own: the provider's error reaches you, the session stays intact, and every prompt can fail until the request shrinks. Try `/compact`.
 

package/dist/src/build-id.d.ts

@@ -1 +1 @@
-export declare const BUILD_ID = "828ac4821689607af006d54a41fd5960bb910bc65aa43adbd81c2f6aac25d180";
+export declare const BUILD_ID = "f637995e7992c76737b37ae27dee5bb6bf47e959fd0f9677c3ae4d0c7530442c";

package/dist/src/build-id.js

@@ -1 +1 @@
-export const BUILD_ID = "828ac4821689607af006d54a41fd5960bb910bc65aa43adbd81c2f6aac25d180";
+export const BUILD_ID = "f637995e7992c76737b37ae27dee5bb6bf47e959fd0f9677c3ae4d0c7530442c";

package/dist/src/compaction.js

@@ -324,6 +324,10 @@ export async function installCompaction(pi, options = {}) {
     // A missing model is configuration, not capacity.
     const missingModel = () => new Error("No context model for the request budget");
     let promptCut;
+    // Where the branch ended when the prompt's cut was planned: the prompt is the entry Pi appends there. A cut that keeps nothing of the conversation
+    // before it names no entry to keep from, and read as it stands that drops the prompt too, so the model never sees it.
+    let promptCutAt = 0;
+    const keptFrom = (cut, ctx) => cut.firstKeptEntryId || ctx.sessionManager.getBranch()[promptCutAt]?.id || "";
     let failure;
     let pendingBoundaryCommit = false;
     // Fresh tool results the room cannot hold, by call id: each request of the run carries the clip in their place; the session keeps them
@@ -523,7 +527,7 @@ export async function installCompaction(pi, options = {}) {
         if (!header)
             throw new Error("Missing native session header for the prompt's cut");
         const preview = SessionManager.inMemory(ctx.cwd, undefined, [header, ...ctx.sessionManager.getBranch()]);
-        preview.appendCompaction(cut.summary, cut.firstKeptEntryId, 0, cut.details, true);
+        preview.appendCompaction(cut.summary, keptFrom(cut, ctx), 0, cut.details, true);
         return preview;
     };
     const withCut = (ctx, cut) => cutSession(ctx, cut).buildSessionProjection().messages;
@@ -533,6 +537,7 @@ export async function installCompaction(pi, options = {}) {
     // Pi's defaults on a 32k window). Each request of the run carries the cut, and the run's first boundary saves it.
     const preflight = async (ctx, incomingPart, observedPrompt, tools) => {
         promptCut = undefined;
+        promptCutAt = ctx.sessionManager.getBranch().length;
         if (!active())
             return;
         access();
@@ -691,7 +696,8 @@ export async function installCompaction(pi, options = {}) {
         const preview = cutSession(ctx, entry), view = Object.create(ctx, { sessionManager: { value: preview } });
         const more = decide && !event.entries.length
             ? boundaryCut({ ...event, context: { ...event.context, contextMessages: preview.buildSessionProjection().messages } }, view, reason) : undefined;
-        const { tokensBefore: _, ...saved } = entry;
+        const { tokensBefore: _, ...cut } = entry;
+        const saved = { ...cut, firstKeptEntryId: keptFrom(entry, ctx) };
         // As for a cut Pi makes (session_compact): the history the cut leaves behind is indexed, once Pi has committed the entries.
         setImmediate(() => { semantic.indexAtCut(semanticProvider(ctx)); });
         return { entries: [...event.entries, saved, ...(more?.entries ?? [])] };

package/dist/src/diagnostics.d.ts

@@ -1,6 +1,6 @@
 import { type RecallPolicy } from './recall.js';
 import type { SessionEntry } from '@earendil-works/pi-coding-agent';
-export declare const DIAGNOSTIC_BUILD = "828ac4821689607af006d54a41fd5960bb910bc65aa43adbd81c2f6aac25d180";
+export declare const DIAGNOSTIC_BUILD = "f637995e7992c76737b37ae27dee5bb6bf47e959fd0f9677c3ae4d0c7530442c";
 export interface Receipt {
     id: string;
     build: string;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-tuck",
-  "version": "1.0.11",
+  "version": "1.0.12",
   "description": "Deterministic compaction for Pi with no model call, plus exact recall of earlier history. Search by meaning on Apple silicon.",
   "keywords": [
     "pi-package",

package/src/build-id.ts

@@ -1 +1 @@
-export const BUILD_ID = "828ac4821689607af006d54a41fd5960bb910bc65aa43adbd81c2f6aac25d180";
+export const BUILD_ID = "f637995e7992c76737b37ae27dee5bb6bf47e959fd0f9677c3ae4d0c7530442c";

package/src/compaction.ts

@@ -304,6 +304,10 @@ export async function installCompaction(pi: ExtensionAPI, options: CompactionOpt
     // before a boundary drops it, and the next prompt decides afresh.
     type CutEntry = { type: "compaction"; summary: string; firstKeptEntryId: string; tokensBefore: number; details?: unknown };
     let promptCut: CutEntry | undefined;
+    // Where the branch ended when the prompt's cut was planned: the prompt is the entry Pi appends there. A cut that keeps nothing of the conversation
+    // before it names no entry to keep from, and read as it stands that drops the prompt too, so the model never sees it.
+    let promptCutAt = 0;
+    const keptFrom = (cut: CutEntry, ctx: ExtensionContext) => cut.firstKeptEntryId || ctx.sessionManager.getBranch()[promptCutAt]?.id || "";
     let failure: Error | undefined;
     let pendingBoundaryCommit = false;
     // Fresh tool results the room cannot hold, by call id: each request of the run carries the clip in their place; the session keeps them
@@ -460,7 +464,7 @@ export async function installCompaction(pi: ExtensionAPI, options: CompactionOpt
         const header = ctx.sessionManager.getHeader();
         if (!header) throw new Error("Missing native session header for the prompt's cut");
         const preview = SessionManager.inMemory(ctx.cwd, undefined, [header, ...ctx.sessionManager.getBranch()]);
-        preview.appendCompaction(cut.summary, cut.firstKeptEntryId, 0, cut.details, true);
+        preview.appendCompaction(cut.summary, keptFrom(cut, ctx), 0, cut.details, true);
         return preview;
     };
     const withCut = (ctx: ExtensionContext, cut: CutEntry) => cutSession(ctx, cut).buildSessionProjection().messages;
@@ -470,6 +474,7 @@ export async function installCompaction(pi: ExtensionAPI, options: CompactionOpt
     // Pi's defaults on a 32k window). Each request of the run carries the cut, and the run's first boundary saves it.
     const preflight = async (ctx: ExtensionContext, incomingPart: number, observedPrompt: string, tools: readonly string[]) => {
         promptCut = undefined;
+        promptCutAt = ctx.sessionManager.getBranch().length;
         if (!active()) return;
         access();
         if (!settingsFor().getCompactionSettings(ctx.model).enabled) return;
@@ -611,7 +616,8 @@ export async function installCompaction(pi: ExtensionAPI, options: CompactionOpt
         const preview = cutSession(ctx, entry), view: ExtensionContext = Object.create(ctx, { sessionManager: { value: preview } });
         const more = decide && !event.entries.length
             ? boundaryCut({ ...event, context: { ...event.context, contextMessages: preview.buildSessionProjection().messages } }, view, reason) : undefined;
-        const { tokensBefore: _, ...saved } = entry;
+        const { tokensBefore: _, ...cut } = entry;
+        const saved = { ...cut, firstKeptEntryId: keptFrom(entry, ctx) };
         // As for a cut Pi makes (session_compact): the history the cut leaves behind is indexed, once Pi has committed the entries.
         setImmediate(() => { semantic.indexAtCut(semanticProvider(ctx)); });
         return { entries: [...event.entries as CutEntry[], saved, ...(more?.entries ?? [])] };