@ngockhoale/ukit 2.7.11 → 2.7.13

package/CHANGELOG.md

@@ -3,6 +3,21 @@
 All notable changes to UKit are documented here.
 
 
+## 2.7.13 - 2026-09-22
+
+- Same content as 2.7.12 (npm staged-publish never completed; re-published as 2.7.13).
+
+## 2.7.12 - 2026-09-22
+
+- **Token & latency diet (cycle C45)**:
+  - Output compression now configurable: `tokenPipeline.outputMaxTokens` (default 600) and `outputMaxLines` (default 40); output already under budget passes through uncompressed — no tee file, no forced re-read.
+  - omp sessions now write memory v2 episodes on session stop (`session-episode.sh` wired into the bridge), so memory recall accumulates under omp.
+  - System prompt ~23% smaller (AGENTS.md 17.1KB → 13.1KB); all rule IDs preserved.
+  - Unattended retry budget reduced to 2 attempts / 2 strategies (UNATTENDED-02).
+  - Removed dead config: `memory.autoCapture`, all `advisor*` keys, `subagents.diffReview*`, `postEditReviewPolicy`, and the code-reviewer `REVIEW_TARGET_TYPE=diff` sidecar mode.
+  - `memory.maxInjectionTokens` now honored within [160, 640] (was silently clamped to 320).
+  - Hook deadline audit: all shipped hooks verified deadline-armed or bounded; watchdog coverage test extended.
+
 ## 2.7.11 - 2026-09-22
 
 - **User permission mode is never overwritten**: `ukit install`/`ukit update`

package/manifests/instructionRules.yaml

@@ -127,8 +127,8 @@ rules:
     title: Bounded attempts and recovery strategies in unattended mode
     kind: supporting
     statement: >-
-      maxAttemptsPerFailure: 5 per failing verification and
-      maxRecoveryStrategies: 3 distinct fix strategies before reassessing —
+      maxAttemptsPerFailure: 2 per failing verification and
+      maxRecoveryStrategies: 2 distinct fix strategies before reassessing —
       do not loop one strategy.
     marker_in: [templates/CLAUDE.md, templates/AGENTS.md, CLAUDE.md, AGENTS.md]
   - id: UNATTENDED-03

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ngockhoale/ukit",
-  "version": "2.7.11",
+  "version": "2.7.13",
   "description": "Install/update an index-first AI workspace for Claude Code, OpenAI Codex and omp (Oh My Pi).",
   "license": "MIT",
   "type": "module",

package/src/core/executionContracts.js

@@ -46,7 +46,6 @@ export const EXECUTION_CONTRACTS = {
     completionRule: 'require-write-and-verification',
     delegationPolicy: 'disallow-by-default',
     completionEvidence: ['write-evidence', 'verification-evidence'],
-    postEditReviewPolicy: 'sidecar-non-blocking',
   },
   'find-cause': {
     maxReadPassesBeforeReassess: 3,
@@ -63,7 +62,6 @@ export const EXECUTION_CONTRACTS = {
     delegationPolicy: 'allow-qualified-sidecar',
     completionEvidence: ['write-evidence', 'verification-evidence'],
     mirrorConsistencyRequired: true,
-    postEditReviewPolicy: 'sidecar-non-blocking',
   },
   'map-impact': {
     maxReadPasses: 3,

package/src/core/output/index.js

@@ -500,7 +500,7 @@ function isTailSummaryLine(line, profile = null) {
   ]);
 }
 
-function buildCompactedSummaryLines(lines, { maxTokens = 180, maxLines = 10, forceFirstCount = 1 } = {}) {
+function buildCompactedSummaryLines(lines, { maxTokens = 600, maxLines = 40, forceFirstCount = 1 } = {}) {
   const hardTokenCap = Math.max(maxTokens, Math.ceil(maxTokens * 2));
   const sourceLines = Array.isArray(lines) ? lines : [];
   const selected = [];
@@ -1007,7 +1007,8 @@ export function compressToolOutput({
   stdout = '',
   stderr = '',
   exitCode = null,
-  maxTokens = 180,
+  maxTokens = 600,
+  maxLines = 40,
   projectRoot = null,
 } = {}) {
   const commandText = String(command ?? '').trim();
@@ -1024,7 +1025,7 @@ export function compressToolOutput({
   const headerLine = `- Recent command: ${displayCommand || 'unknown command'}${exitCode === null || exitCode === undefined ? '' : ` (exit ${exitCode})`}`;
   const compactedLines = buildCompactedSummaryLines([headerLine, ...candidateLines.map((line) => `- ${line}`)], {
     maxTokens,
-    maxLines: 10,
+    maxLines,
   });
   let summary = compactedLines.join('\n').trim();
   let validationMode = 'default';
@@ -1037,7 +1038,7 @@ export function compressToolOutput({
       ...candidateLines.map((line) => `- ${line}`),
     ], {
       maxTokens,
-      maxLines: 10,
+      maxLines,
       forceFirstCount: Math.min(1 + anchorLines.length, 5),
     });
     const anchorFirstSummary = anchorFirstLines.join('\n').trim();
@@ -1052,7 +1053,7 @@ export function compressToolOutput({
         ...anchorLines.map((line) => `- ${line}`),
       ], {
         maxTokens,
-        maxLines: 10,
+        maxLines,
         forceFirstCount: Math.min(1 + anchorLines.length, 5),
       }).join('\n').trim();
       validationMode = 'forced-anchors';
@@ -1236,7 +1237,8 @@ export async function captureCompressedToolOutput(
     stderr = '',
     exitCode = null,
     promptCache = true,
-    maxTokens = 180,
+    maxTokens = 600,
+    maxLines = 40,
   } = {},
 ) {
   const requestKey = buildCompactMachineKey('tool-output-v1', {
@@ -1245,6 +1247,7 @@ export async function captureCompressedToolOutput(
     stderrHash: buildCompactMachineKey('stderr', stripAnsi(stderr)),
     exitCode,
     maxTokens,
+    maxLines,
   });
 
   if (promptCache) {
@@ -1293,6 +1296,7 @@ export async function captureCompressedToolOutput(
     stderr,
     exitCode,
     maxTokens,
+    maxLines,
     projectRoot,
   });
   const recoveryReason = buildRawOutputRecoveryReason({

package/src/core/runtimeConfig.js

@@ -130,18 +130,16 @@ export function buildDefaultRuntimeConfig(overrides = {}) {
       inputCompression: true,
       outputCompression: true,
       promptCache: true,
+      outputMaxTokens: 600,
+      outputMaxLines: 40,
     },
     router: {
       enabled: true,
       defaultModel: 'claude-sonnet-5',
-      advisorModel: 'claude-opus-5',
-      advisorEnabled: true,
-      maxAdvisorCalls: 3,
     },
     orchestration: {
       enabled: true,
       orchestratorModel: 'claude-sonnet-5',
-      advisorEnabled: true,
       permissionMode: 'unattended',
       // FR-003: eval-subprocess gate for rendered omp approval (tools.approval.eval).
       // Fail-closed: only an explicit user `true` renders `eval: allow`; absent,
@@ -207,9 +205,8 @@ export function buildDefaultRuntimeConfig(overrides = {}) {
     },
     memory: {
       enabled: true,
-      autoCapture: true,
       progressiveRetrieval: true,
-      maxInjectionTokens: 1000,
+      maxInjectionTokens: 320,
       archiveAfterDays: 30,
       maxSessions: 20,
       maxArchivedSessions: 50,
@@ -245,9 +242,6 @@ export function buildDefaultRuntimeConfig(overrides = {}) {
       enabled: true,
       smallTaskModel: 'unic-lite',
       smallTaskAgent: 'ukit-small-task-maintainer',
-      diffReviewEnabled: true,
-      diffReviewAgent: 'code-reviewer',
-      diffReviewModel: 'unic-smart',
       visionEnabled: true,
       visionModel: 'unic-vision',
       visionAgent: 'ukit-vision-analyst',
@@ -383,6 +377,14 @@ export function validateRuntimeConfig(config) {
     pushBooleanError(errors, config.tokenPipeline.inputCompression, 'tokenPipeline.inputCompression');
     pushBooleanError(errors, config.tokenPipeline.outputCompression, 'tokenPipeline.outputCompression');
     pushBooleanError(errors, config.tokenPipeline.promptCache, 'tokenPipeline.promptCache');
+    // FR-001: optional-present — absent keys fall back to code defaults on merge;
+    // when present they must be positive numbers.
+    if (config.tokenPipeline.outputMaxTokens !== undefined) {
+      pushPositiveNumberError(errors, config.tokenPipeline.outputMaxTokens, 'tokenPipeline.outputMaxTokens');
+    }
+    if (config.tokenPipeline.outputMaxLines !== undefined) {
+      pushPositiveNumberError(errors, config.tokenPipeline.outputMaxLines, 'tokenPipeline.outputMaxLines');
+    }
   }
 
   if (!isPlainObject(config.router)) {
@@ -392,11 +394,6 @@ export function validateRuntimeConfig(config) {
     if (typeof config.router.defaultModel !== 'string' || config.router.defaultModel.trim() === '') {
       errors.push('router.defaultModel must be a non-empty string.');
     }
-    if (typeof config.router.advisorModel !== 'string' || config.router.advisorModel.trim() === '') {
-      errors.push('router.advisorModel must be a non-empty string.');
-    }
-    pushBooleanError(errors, config.router.advisorEnabled, 'router.advisorEnabled');
-    pushPositiveNumberError(errors, config.router.maxAdvisorCalls, 'router.maxAdvisorCalls');
   }
 
   if (!isPlainObject(config.orchestration)) {
@@ -404,7 +401,6 @@ export function validateRuntimeConfig(config) {
   } else {
     pushBooleanError(errors, config.orchestration.enabled, 'orchestration.enabled');
     pushNonEmptyStringError(errors, config.orchestration.orchestratorModel, 'orchestration.orchestratorModel');
-    pushBooleanError(errors, config.orchestration.advisorEnabled, 'orchestration.advisorEnabled');
     // permissionMode is optional-present: absent → tolerated (pre-C35 configs);
     // present → must be a reserved mode.
     if (config.orchestration.permissionMode !== undefined
@@ -560,7 +556,6 @@ export function validateRuntimeConfig(config) {
     errors.push('memory must be an object.');
   } else {
     pushBooleanError(errors, config.memory.enabled, 'memory.enabled');
-    pushBooleanError(errors, config.memory.autoCapture, 'memory.autoCapture');
     pushBooleanError(errors, config.memory.progressiveRetrieval, 'memory.progressiveRetrieval');
     pushPositiveNumberError(errors, config.memory.maxInjectionTokens, 'memory.maxInjectionTokens');
     pushPositiveNumberError(errors, config.memory.archiveAfterDays, 'memory.archiveAfterDays');

package/src/core/status.js

@@ -201,7 +201,6 @@ export async function buildStatusReport(projectRoot) {
       (entry) => `${entry.profile} ${entry.entryCount} / ${entry.savingsLabel}`,
     ),
     defaultModel: config.router.defaultModel,
-    advisorEnabled: Boolean(config.router.advisorEnabled),
     projectImportant,
   };
 }
@@ -251,6 +250,6 @@ export function formatStatusReport(report) {
     `${padLabel('Cache lanes')} ${report.cacheLanes}`,
     `${padLabel('Output comp.')} ${report.outputCompression}`,
     `${padLabel('Output lanes')} ${report.outputLanes}`,
-    `${padLabel('Model')} ${report.defaultModel} (advisor: ${report.advisorEnabled ? 'on' : 'off'})`,
+    `${padLabel('Model')} ${report.defaultModel}`,
   ];
 }

package/src/index/taskRouting.js

@@ -271,9 +271,6 @@ export function buildRouteSummary({
   });
   const executionContract = buildExecutionContract(executionMode);
   const continuousExecution = buildContinuousExecutionPolicy(autonomyLevel);
-  const postEditReview = executionContract?.postEditReviewPolicy
-    ? { policy: executionContract.postEditReviewPolicy, agent: 'code-reviewer', reviewTargetType: 'diff' }
-    : null;
   const completionState = buildCompletionState({
     executionMode,
     verificationRecommendation,
@@ -311,7 +308,6 @@ export function buildRouteSummary({
     editGuardHint ? `editGuard=${editGuardHint}` : null,
     delegationRecommendation?.hint ? `delegate=${delegationRecommendation.hint}` : null,
     policyMode ? `policy=${policyMode}` : null,
-    postEditReview ? `review=${postEditReview.agent}(${postEditReview.reviewTargetType})` : null,
     handoffBudget?.warning ? `budget=${handoffBudget.warning}` : null,
     worklogBudget?.warning ? `budget=${worklogBudget.warning}` : null,
   ].filter(Boolean).join(' | ');
@@ -328,7 +324,6 @@ export function buildRouteSummary({
     approachSelector,
     executionContract,
     completionState,
-    postEditReview,
     continuationState,
     autonomyLevel,
     continuousExecution,

package/templates/.claude/agents/code-reviewer.md

@@ -1,6 +1,6 @@
 ---
 name: code-reviewer
-description: "Independent reviewer for handoff Phase 3, for spec/plan documents, and for non-blocking sidecar diff review of daily-flow edits. For code (default): use after executor reports STATUS: DONE on a handoff task, MUST run with a model different from the executor (configured in .ukit/storage/config.json → handoff.reviewer.model, default unic-smart), produces a verdict: APPROVED | APPROVED-WITH-MINOR | CHANGES-REQUESTED | CRITICAL. For spec/plan documents (set REVIEW_TARGET_TYPE=spec or plan): reviews a docs/plans/*.md file for completeness/consistency/clarity/scope/YAGNI, produces Status: Approved | Issues Found. For sidecar diff review (set REVIEW_TARGET_TYPE=diff): reviews the current uncommitted git diff after a local-build/shared-edit task, runs in the background and never blocks the main task, produces STATUS: clean | issues-found."
+description: "Independent reviewer for handoff Phase 3 and for spec/plan documents. For code (default): use after executor reports STATUS: DONE on a handoff task, MUST run with a model different from the executor (configured in .ukit/storage/config.json → handoff.reviewer.model, default unic-smart), produces a verdict: APPROVED | APPROVED-WITH-MINOR | CHANGES-REQUESTED | CRITICAL. For spec/plan documents (set REVIEW_TARGET_TYPE=spec or plan): reviews a docs/plans/*.md file for completeness/consistency/clarity/scope/YAGNI, produces Status: Approved | Issues Found."
 model: opus  # unic-smart
 color: yellow
 tools: ["Read", "Grep", "Glob", "Bash"]
@@ -14,7 +14,6 @@ You are the independent reviewer for UKit's handoff Quality Gate. Your model is
 
 - `code` (default, if not specified) — reviewing a handoff task diff. Follow **Code Review** below, unchanged.
 - `spec` | `plan` — reviewing a document (e.g. `docs/plans/*.md`), no diff/task file/executor report involved. Skip straight to **Spec/Plan Review** at the end of this file instead.
-- `diff` — non-blocking sidecar review of the current uncommitted diff in the daily (non-handoff) flow. No task file/executor report/model-isolation check involved. Skip straight to **Sidecar Diff Review** at the end of this file instead.
 
 ## Code Review (REVIEW_TARGET_TYPE=code)
 
@@ -175,43 +174,3 @@ NOTES: [1-2 sentences if needed]
 
 `<N>` = 1 + however many `### Round` entries already exist in the log (1 if this is the first review).
 
-## Sidecar Diff Review (REVIEW_TARGET_TYPE=diff)
-
-This mode exists so a weaker daily-flow executor model still gets a second pair of eyes,
-without adding wait time to the main task. You are launched in the background right after
-the main task already has write + verification evidence; the caller is not waiting on you.
-
-### Inputs you expect
-
-- No task file, no executor report, no model-isolation check. Just read the current uncommitted
-  diff yourself: `git diff` (and `git diff --stat` for an overview). If there is no diff, report
-  `STATUS: clean` with `FINDINGS: none` and stop.
-
-### Review order
-
-Apply the same lenses as Code Review's steps 2-7, scoped to what the diff actually touches:
-
-1. **Correctness** — Does the diff do what it looks like it's trying to do? Wrong assumptions, stale refs, missing cases?
-2. **Regression risk** — Any existing behavior/tests/contracts this plausibly breaks?
-3. **Safety / security / data loss** — Destructive actions, auth/permission, path handling, unsafe shell/DB/file ops.
-4. **Performance / scale** — Accidental N+1, repeated I/O, large scans in hot paths.
-5. **Maintainability** — Duplicated logic, dead branches, misleading naming, drift between docs/tests/source.
-6. **Solution fit** — Apply the same 5 evidence-first questions as Code Review step 7 (duplicate semantics / standard library or native suffices / simplification dropping validation, security, accessibility or tests / shared root cause / speculative abstraction). **Never grade brevity or line count as a quality win.** Findings must cite the construct + reason, and here they remain advisory hypotheses only — never a delete-list, never an edit.
-
-Do not re-run the project's full verification suite here — this is an advisory pass, not a gate.
-You may read files for context but this mode never edits anything.
-
-### Output
-
-Keep it short — this is a quick advisory pass, not a full verdict:
-
-```
-STATUS: clean | issues-found
-FINDINGS:
-  - file:line — what's wrong, why it matters
-NOTES: [advisory only, non-blocking — 1 sentence if needed]
-```
-
-There is no task file or INDEX.md to update in this mode. Findings are advisory only: the main
-task is not blocked on this review and may already be reported done by the time you finish.
-Report back to the caller in a few lines; do not paste the full diff.

package/templates/.claude/ukit/index/route-task.mjs

@@ -2683,7 +2683,6 @@ function buildExecutionContract(executionMode = null) {
       completionRule: 'require-write-and-verification',
       delegationPolicy: 'disallow-by-default',
       completionEvidence: ['write-evidence', 'verification-evidence'],
-      postEditReviewPolicy: 'sidecar-non-blocking',
     },
     'find-cause': {
       modelTier: 'code',
@@ -2702,7 +2701,6 @@ function buildExecutionContract(executionMode = null) {
       delegationPolicy: 'allow-qualified-sidecar',
       completionEvidence: ['write-evidence', 'verification-evidence'],
       mirrorConsistencyRequired: true,
-      postEditReviewPolicy: 'sidecar-non-blocking',
     },
     'map-impact': {
       modelTier: 'code',

package/templates/.claude/ukit/runtime/output-compression.mjs

@@ -60,7 +60,8 @@ if (Number.isFinite(HOOK_DEADLINE_MS) && HOOK_DEADLINE_MS > 0) {
 }
 
 const ANSI_RE = /\u001b\[[0-9;]*m/g;
-const DEFAULT_MAX_OUTPUT_TOKENS = 180;
+const DEFAULT_MAX_OUTPUT_TOKENS = 600;
+const DEFAULT_MAX_OUTPUT_LINES = 40;
 const DEFAULT_OUTPUT_HISTORY_MAX_ENTRIES = 25;
 const RAW_OUTPUT_SAVE_MIN_TOKENS = 600;
 const RAW_OUTPUT_SAVE_MIN_SAVED_TOKENS = 250;
@@ -582,7 +583,7 @@ function isTailSummaryLine(line, profile = null) {
   ]);
 }
 
-function buildCompactedSummaryLines(lines, { maxTokens = DEFAULT_MAX_OUTPUT_TOKENS, maxLines = 10, forceFirstCount = 1 } = {}) {
+function buildCompactedSummaryLines(lines, { maxTokens = DEFAULT_MAX_OUTPUT_TOKENS, maxLines = DEFAULT_MAX_OUTPUT_LINES, forceFirstCount = 1 } = {}) {
   const sourceLines = Array.isArray(lines) ? lines : [];
   const selected = [];
   const seen = new Set();
@@ -1089,6 +1090,7 @@ function compressToolOutput({
   stderr = '',
   exitCode = null,
   maxTokens = DEFAULT_MAX_OUTPUT_TOKENS,
+  maxLines = DEFAULT_MAX_OUTPUT_LINES,
   projectRoot = null,
 } = {}) {
   const commandText = String(command ?? '').trim();
@@ -1105,7 +1107,7 @@ function compressToolOutput({
   const headerLine = `- Recent command: ${displayCommand || 'unknown command'}${exitCode === null || exitCode === undefined ? '' : ` (exit ${exitCode})`}`;
   const compactedLines = buildCompactedSummaryLines([headerLine, ...candidateLines.map((line) => `- ${line}`)], {
     maxTokens,
-    maxLines: 10,
+    maxLines,
   });
   let summary = compactedLines.join('\n').trim();
   const missingAnchors = findMissingAnchors(summary, anchorLines);
@@ -1117,7 +1119,7 @@ function compressToolOutput({
       ...candidateLines.map((line) => `- ${line}`),
     ], {
       maxTokens,
-      maxLines: 10,
+      maxLines,
       forceFirstCount: Math.min(1 + anchorLines.length, 5),
     }).join('\n').trim();
 
@@ -1130,7 +1132,7 @@ function compressToolOutput({
         ...anchorLines.map((line) => `- ${line}`),
       ], {
         maxTokens,
-        maxLines: 10,
+        maxLines,
         forceFirstCount: Math.min(1 + anchorLines.length, 5),
       }).join('\n').trim();
     }
@@ -1317,10 +1319,19 @@ async function loadRuntimeConfig(projectRoot) {
     tokenPipeline: {
       outputCompression: true,
       promptCache: true,
+      outputMaxTokens: DEFAULT_MAX_OUTPUT_TOKENS,
+      outputMaxLines: DEFAULT_MAX_OUTPUT_LINES,
     },
   }, await readJson(runtimePaths.configPath, {}));
 }
 
+// FR-001: the hook's local merge does not run schema validation — a malformed
+// value ("abc", -5) must fall back to the default, never throw or disable the
+// skip-small fast path.
+function positiveNumberOrDefault(value, fallback) {
+  return (typeof value === 'number' && Number.isFinite(value) && value > 0) ? value : fallback;
+}
+
 async function main() {
   const rawInput = await readStdin();
   const payload = (() => {
@@ -1360,11 +1371,26 @@ async function main() {
     return;
   }
 
+  // FR-003 skip-small fast path: when the raw output already fits the configured
+  // budget, exit before the prompt-cache lookup, tee write, and history append —
+  // compressing it would cost a re-read that outweighs the savings.
+  const outputMaxTokens = positiveNumberOrDefault(
+    config?.tokenPipeline?.outputMaxTokens, DEFAULT_MAX_OUTPUT_TOKENS);
+  const outputMaxLines = positiveNumberOrDefault(
+    config?.tokenPipeline?.outputMaxLines, DEFAULT_MAX_OUTPUT_LINES);
+  const rawText = [command, stdout, stderr].filter(Boolean).join('\n');
+  if (estimateTokenCount(rawText) <= outputMaxTokens) {
+    process.exit(0);
+    return;
+  }
+
   const requestKey = buildCompactMachineKey('tool-output-v1', {
     command,
     stdoutHash: buildCompactMachineKey('stdout', stripAnsi(stdout)),
     stderrHash: buildCompactMachineKey('stderr', stripAnsi(stderr)),
     exitCode,
+    maxTokens: outputMaxTokens,
+    maxLines: outputMaxLines,
   });
 
   if (config?.tokenPipeline?.promptCache) {
@@ -1410,6 +1436,8 @@ async function main() {
     stdout,
     stderr,
     exitCode,
+    maxTokens: outputMaxTokens,
+    maxLines: outputMaxLines,
     projectRoot,
   });
 

package/templates/.claude/ukit/runtime/reinject-context.mjs

@@ -143,7 +143,7 @@ async function main() {
     || line.startsWith('- Recent command output:')
     || line.startsWith('- Recent delegation hint:')
   ));
-  const maxTokens = Math.max(160, Math.min(320, Number(config.memory?.maxInjectionTokens) || 320));
+  const maxTokens = Math.max(160, Math.min(640, Number(config.memory?.maxInjectionTokens) || 320));
   const thresholdPlan = await buildThresholdCompactPlan({
     state: pressureState,
     config,

package/templates/.codex/settings.json

@@ -262,7 +262,6 @@
       "enabled": true,
       "configPath": ".ukit/storage/config.json",
       "modelField": "orchestration.orchestratorModel",
-      "advisorField": "orchestration.advisorEnabled",
       "defaultModel": "claude-sonnet-5",
       "internalOnly": true,
       "qualityFirst": true,

package/templates/.omp/agents/code-reviewer.md

@@ -1,6 +1,6 @@
 ---
 name: code-reviewer
-description: "Independent reviewer for handoff Phase 3, for spec/plan documents, and for non-blocking sidecar diff review of daily-flow edits. For code (default): use after executor reports STATUS: DONE on a handoff task, MUST run with a model different from the executor (configured in .ukit/storage/config.json → handoff.reviewer.model, default unic-smart), produces a verdict: APPROVED | APPROVED-WITH-MINOR | CHANGES-REQUESTED | CRITICAL. For spec/plan documents (set REVIEW_TARGET_TYPE=spec or plan): reviews a docs/plans/*.md file for completeness/consistency/clarity/scope/YAGNI, produces Status: Approved | Issues Found. For sidecar diff review (set REVIEW_TARGET_TYPE=diff): reviews the current uncommitted git diff after a local-build/shared-edit task, runs in the background and never blocks the main task, produces STATUS: clean | issues-found."
+description: "Independent reviewer for handoff Phase 3 and for spec/plan documents. For code (default): use after executor reports STATUS: DONE on a handoff task, MUST run with a model different from the executor (configured in .ukit/storage/config.json → handoff.reviewer.model, default unic-smart), produces a verdict: APPROVED | APPROVED-WITH-MINOR | CHANGES-REQUESTED | CRITICAL. For spec/plan documents (set REVIEW_TARGET_TYPE=spec or plan): reviews a docs/plans/*.md file for completeness/consistency/clarity/scope/YAGNI, produces Status: Approved | Issues Found."
 model: "@slow"
 tools: ["read","grep","glob","bash"]
 ---
@@ -13,7 +13,6 @@ You are the independent reviewer for UKit's handoff Quality Gate. Your model is
 
 - `code` (default, if not specified) — reviewing a handoff task diff. Follow **Code Review** below, unchanged.
 - `spec` | `plan` — reviewing a document (e.g. `docs/plans/*.md`), no diff/task file/executor report involved. Skip straight to **Spec/Plan Review** at the end of this file instead.
-- `diff` — non-blocking sidecar review of the current uncommitted diff in the daily (non-handoff) flow. No task file/executor report/model-isolation check involved. Skip straight to **Sidecar Diff Review** at the end of this file instead.
 
 ## Code Review (REVIEW_TARGET_TYPE=code)
 
@@ -174,43 +173,3 @@ NOTES: [1-2 sentences if needed]
 
 `<N>` = 1 + however many `### Round` entries already exist in the log (1 if this is the first review).
 
-## Sidecar Diff Review (REVIEW_TARGET_TYPE=diff)
-
-This mode exists so a weaker daily-flow executor model still gets a second pair of eyes,
-without adding wait time to the main task. You are launched in the background right after
-the main task already has write + verification evidence; the caller is not waiting on you.
-
-### Inputs you expect
-
-- No task file, no executor report, no model-isolation check. Just read the current uncommitted
-  diff yourself: `git diff` (and `git diff --stat` for an overview). If there is no diff, report
-  `STATUS: clean` with `FINDINGS: none` and stop.
-
-### Review order
-
-Apply the same lenses as Code Review's steps 2-7, scoped to what the diff actually touches:
-
-1. **Correctness** — Does the diff do what it looks like it's trying to do? Wrong assumptions, stale refs, missing cases?
-2. **Regression risk** — Any existing behavior/tests/contracts this plausibly breaks?
-3. **Safety / security / data loss** — Destructive actions, auth/permission, path handling, unsafe shell/DB/file ops.
-4. **Performance / scale** — Accidental N+1, repeated I/O, large scans in hot paths.
-5. **Maintainability** — Duplicated logic, dead branches, misleading naming, drift between docs/tests/source.
-6. **Solution fit** — Apply the same 5 evidence-first questions as Code Review step 7 (duplicate semantics / standard library or native suffices / simplification dropping validation, security, accessibility or tests / shared root cause / speculative abstraction). **Never grade brevity or line count as a quality win.** Findings must cite the construct + reason, and here they remain advisory hypotheses only — never a delete-list, never an edit.
-
-Do not re-run the project's full verification suite here — this is an advisory pass, not a gate.
-You may read files for context but this mode never edits anything.
-
-### Output
-
-Keep it short — this is a quick advisory pass, not a full verdict:
-
-```
-STATUS: clean | issues-found
-FINDINGS:
-  - file:line — what's wrong, why it matters
-NOTES: [advisory only, non-blocking — 1 sentence if needed]
-```
-
-There is no task file or INDEX.md to update in this mode. Findings are advisory only: the main
-task is not blocked on this review and may already be reported done by the time you finish.
-Report back to the caller in a few lines; do not paste the full diff.

package/templates/.omp/hooks/pre/ukit-bridge.js

@@ -64,6 +64,11 @@ export const HOOK_EVENT_MAP = {
   before_agent_start: ['sensitive-data-guard.mjs', 'skill-router.sh', 'vision-router.sh', 'context-window-guard.sh'],
   'session.compacting': ['reinject-context.sh'],
   session_start: ['project-important.sh', 'auto-prune-bash.sh', 'reset-compact-pressure.sh', 'handoff-resume.sh'],
+  // TASK-002 (cycle 45): omp emits `session_stop` (no `session_end`); the chain
+  // runs only on RELEASED stops inside runSessionStop — never on blocked or
+  // dedupe-skipped stops, since `ukit memory episode` dedupes on meta.ledgerKey
+  // and a partial episode would suppress the real one.
+  session_stop: ['session-episode.sh'],
 };
 
 const TOOL_NAME_MAP = {
@@ -129,6 +134,7 @@ export const ADVISORY_SCRIPTS = new Set([
   'auto-prune-bash.sh',
   'reset-compact-pressure.sh',
   'handoff-resume.sh',
+  'session-episode.sh',
 ]);
 
 function classifyFailure(scriptName) {
@@ -971,6 +977,19 @@ function loadStopCoordinatorModule() {
   }
   return stopCoordinatorModulePromise;
 }
+// TASK-002 (cycle 45): the session-episode chain is advisory end-to-end — a
+// failing script (or a transport failure) warns and never changes the stop
+// decision. Called ONLY on released stops: a blocked stop means the turn
+// continues, and `ukit memory episode` dedupes on meta.ledgerKey, so firing
+// early would freeze a partial episode and suppress the real one.
+async function runSessionEpisodeChain(pi, payload, projectRoot) {
+  try {
+    await runScriptChain(pi, HOOK_EVENT_MAP.session_stop, payload, { projectRoot });
+  } catch (error) {
+    pi.logger?.warn?.(`[UKit] session-episode chain failed (advisory): ${error?.message || error}`);
+  }
+}
+
 
 export async function runSessionStop(
   pi,
@@ -1061,6 +1080,7 @@ export async function runSessionStop(
         + 'The completion gate is DOWN, not your work: run `ukit install` to repair, '
         + 'then re-send the task in a new message if the run must continue.',
       ], 'nextTurn', { display: true });
+      await runSessionEpisodeChain(pi, payload, projectRoot);
       return undefined;
     }
     pi.logger?.warn?.(`[UKit] completion evaluator crashed (streak ${streak}/${maxStreak}): ${error?.message || error}`);
@@ -1099,6 +1119,7 @@ export async function runSessionStop(
     if (handoffBlock) {
       return { continue: true, additionalContext: handoffBlock };
     }
+    await runSessionEpisodeChain(pi, payload, projectRoot);
     return undefined;
   }