@lmzhen/dsh-evolution-review 0.3.73 → 0.3.76

package/README.md

@@ -21,6 +21,7 @@ Independent of request-prefix construction. This package does not alter the asse
 ## Known Limitations and Deferred Work
 
 
+- The review's model/context routing below applies to `reviewMode: 'subagent'` (opt-in since 0.3.74); in the default `'inject'` mode the review runs on the parent agent's model inside the parent thread.
 - Review subagents are spawned with the plain `skill` tool only (`reviewToolAllow` default and the host/preset config both = `[skill]` — the DSH tool catalog has no `skill_search`/`skill_load` discovery pair, so the Hermes-lineage Anchored Standard `skill_search`/`skill_load` allow-list does not exist here).
 - Review subagents run as `spawn` children on the deployment default preset rather than inheriting the parent agent's composition (`fork`): a fork child is always promoted by the Anchored Standard bootstrap and its narrowed resident catalog would drop the plain `skill` tool from the review allow-list.
 - The review request text is redacted for credential-shaped patterns before it reaches the subagent, but redaction is pattern-based and best-effort, not a security boundary.
@@ -29,6 +30,8 @@ Independent of request-prefix construction. This package does not alter the asse
 - `evolution/review-scheduled` and `evolution/review-error` are emitted for platform/user wiring only — this family has no in-repo production `ctx.on` consumer for them. They are declared externally owned (the platform side wires consumption), which matches the `EXEMPT_ORPHANS` set in `scripts/verify-event-pairing.mjs`.
 - When the `evolution-state` service is not mounted, the memory/skill cadence state is not persisted and every turn restarts from a clean `{ turnsSinceMemory: 0, turnsSinceSkill: 0 }` baseline — the review schedule is stateless and re-decided each turn rather than accumulating across the conversation. The loss is surfaced once per process as a logger warning at the first turn/end.
 - Read-before-write can see the review subagent's own `skill` reads only when the subagent backend exposes `localAgent` (the in-process driver does; out-of-process backends such as ACP and the CLI providers set `localAgent: undefined`). With a remote backend the subagent's reads are invisible, so a plan item patching a skill the subagent itself loaded is dropped as "unread" — the review then falls back to the parent session's reads only. Documented rather than worked around: recovering the child read set needs a `SubagentLike` contract change (v14 P2-6).
+- **The default `reviewMode: 'inject'` produces no plan ledger.** The review runs in the parent session and emits no `evolution/plan-applied` event, so `evolution-activity`'s `activity.json` and the `evolution-replay` leaderboard never grow in this mode — the only production emit point sits inside the subagent path. Set `reviewMode: 'subagent'` on the `evolution-policy` row when the plan audit trail is required; the plugin states this once at load instead of leaving an empty ledger to be read as "no reviews happened" (v37 S2.2, plan decision (c)).
+- **`.pinned` protection in the `'inject'` channel rides a family-internal session mark.** The review prompt marks the parent session, `tool-skill-manage` reads that mark and resolves both origin surfaces (approval + library) to `background_review`, and the next REAL user message (`source.kind === 'user'`) clears it — plugin notices, including the review prompt itself, do not. Two bounds follow from the platform's inject contract (no driver wake; a prompt is dropped on cancel/dispose and may be missed with an already-claimed batch): with `reviewWakeInject: false`, or on a host without `followup`, (a) a pending prompt may never execute while the session stays idle, and (b) when the user's next message wakes the session, the mark is cleared before that pending prompt reaches the model — its writes are then attributed `foreground` and the pinned guard does not cover them. The mark is in-memory, so a restart drops the window (v37 S2.2, plan decision (b)).
 
 ## Configuration
 
@@ -38,7 +41,9 @@ Independent of request-prefix construction. This package does not alter the asse
 
 ### Review delivery contract (0.3.38-0.3.42)
 
-- **Both channels execute at conversation end only** (a `turn/end` with `reason.kind === 'completed'`): a cadence threshold fire mid-task merely latches the kind — no subagent spawn, no inject. The flush runs BEFORE the latch block (the completing turn may itself be a threshold-firing turn). `reviewMode` (`'subagent'` default / `'inject'`) selects how the flush delivers; the explicit `'inject'` mode's historical "immediate on threshold" contract was superseded in 0.3.39 — both modes are end-of-conversation.
+- **Both channels execute at conversation end only** (a `turn/end` with `reason.kind === 'completed'`): a cadence threshold fire mid-task merely latches the kind — no subagent spawn, no inject. The flush runs BEFORE the latch block (the completing turn may itself be a threshold-firing turn). `reviewMode` selects how the flush delivers: **`'inject'` is the default since 0.3.74**, `'subagent'` is an explicit opt-in. Rationale: the parent session already holds a warm prefix cache, so an injected prompt costs the new tokens only, while a spawned child re-prefills its own system prompt plus a redacted, re-serialized conversation digest (`buildReviewRequest`) under a **different model** (`skillReviewModel`/`memoryReviewModel`) — no prefix is shared with the parent, so the whole child input is paid at full price. `'subagent'` remains the choice for deployments that want the parent context kept clean (a review's skill reads and plan do not join the parent thread) or a dedicated review model; the subagent-only knobs (`reviewProvider`, `reviewTimeoutMs`, `reviewMaxDepth`, `reviewToolAllow`, the review models) are inert in inject mode. The explicit `'inject'` mode's historical "immediate on threshold" contract was superseded in 0.3.39 — both modes are end-of-conversation.
 - **`skillReviewTrigger`** (default `'cadence'`): the cadence channel is **always on** (one end-of-conversation review from the cadence latch per task segment); the flag gates **only the completion channel** — `'cadence'` disables it, `'completion'` enables it (cadence still fires), `'both'` enables it on top of the always-on cadence. At one boundary a turn is served by exactly one review: the cadence flush runs first and returns, so `'both'` never double-sends a second task-complete prompt at the same boundary (V10-13).
-- **`reviewWakeInject`** (default `true`): deliveries use `agent.followup` (next-turn + wake — the model starts processing immediately) instead of the non-waking `agent.inject` (which waits for the next driver wake). The host falls back to `inject` when it has no followup or the option is `false`. **The wake primitive is always called ON the agent instance** — the platform's `Agent.followup`/`inject` are prototype methods that call `this.send(...)`, so extracting one into a local and calling the detached reference throws (0.3.73: that throw was caught and logged while the cadence reset still ran, silently consuming every segment's review from 2026-09-07). A refused delivery now returns `false` and the caller keeps its latch and counters, so the review retries at the next completed boundary instead of vanishing; `evolution-host/tests/wake-delivery-guard.spec.ts` pins the call form mechanically. The woken turn's own cadence fire is suppressed once (an injected review prompt alone must not re-trigger a review under `interval=1`); a restart clears the queue, so the loop cannot survive it.
+- **`reviewWakeInject`** (default `true`): deliveries use `agent.followup` (next-turn + wake — the model starts processing immediately) instead of the non-waking `agent.inject` (which waits for the next driver wake). The host falls back to `inject` when it has no followup or the option is `false`. **The wake primitive is always called ON the agent instance** — the platform's `Agent.followup`/`inject` are prototype methods that call `this.send(...)`, so extracting one into a local and calling the detached reference throws (0.3.73: that throw was caught and logged while the cadence reset still ran, silently consuming every segment's review from 2026-09-07). A refused delivery now returns `false` and the caller keeps its latch and counters, so the review retries at the next completed boundary instead of vanishing; rule N13b in `packages/scripts/verify-arch-guards.mjs` pins the call form mechanically (comments and string literals are masked, and the detector self-tests at startup). The woken turn's own cadence fire is suppressed once (an injected review prompt alone must not re-trigger a review under `interval=1`); a restart clears the queue, so the loop cannot survive it.
 - **Counting window = injection-to-injection**: the `turnsSinceMemory`/`turnsSinceSkill` counters are monotonic across threshold fires (`resetOnFire: false`) and are zeroed at the flush delivery — a continued conversation starts a fresh segment from the injection. A threshold fire on the completing turn is caught by the flush (`pendingKind = latch ?? kind`). All deliveries (review prompt AND result notices) share the same waking channel; a failed counter-reset persist warns once per session (a stateful reload may re-deliver).
+
+**Runtime invariant:** No companion is published. The platform auto-assembles nothing and the family mounts no `<pkg>/invariant` cordis row, so a companion here would never execute (v37 S2.1 / I-3).

package/lib/index.js

@@ -1,7 +1,8 @@
 import { createHash, randomUUID } from "node:crypto";
 import z from "@deepseek-ai/schemastery";
 import { createUserMessage } from "@deepseek-ai/dsh-llm";
-import { COMPLETION_SKILL_REVIEW_PROMPT, DEFAULT_MAX_OPS_PER_PLAN, DEFAULT_MEMORY_CHAR_LIMIT, DEFAULT_MEMORY_REVIEW_MODEL, DEFAULT_REVIEW_CONTEXT_MESSAGES, DEFAULT_REVIEW_MEMORY_INTERVAL, DEFAULT_REVIEW_MESSAGE_CHARS, DEFAULT_REVIEW_SKILL_INTERVAL, DEFAULT_REVIEW_TIMEOUT_MS, DEFAULT_SKILL_CONTENT_CHARS, DEFAULT_SKILL_LIMITS, DEFAULT_SKILL_REVIEW_COMPLETION_MIN_TOOL_CALLS, DEFAULT_SKILL_REVIEW_MODEL, DEFAULT_SKILL_REVIEW_TRIGGER, DEFAULT_SUBSTANTIVE_MIN_AGENT_CHARS, DEFAULT_SUBSTANTIVE_MIN_TOOL_CALLS, DEFAULT_SUBSTANTIVE_MIN_USER_CHARS, DEFAULT_USER_CHAR_LIMIT, MAX_TIMER_DELAY_MS, PROMPT_BUNDLE, SkillLibrary, advanceReview, assertSkillsRootAliasRetired, clampedNumber, contentHash, evolutionIoAdapter, foldTurn, redactSecrets, resolveOrigins, resolveRootConfig, resolveSkillsRoot, reviewPrompt, verifyPromptBundle } from "@lmzhen/dsh-evolution-core";
+import { SessionId } from "@deepseek-ai/dsh-session";
+import { COMPLETION_SKILL_REVIEW_PROMPT, DEFAULT_MAX_OPS_PER_PLAN, DEFAULT_MEMORY_CHAR_LIMIT, DEFAULT_MEMORY_REVIEW_MODEL, DEFAULT_REVIEW_CONTEXT_MESSAGES, DEFAULT_REVIEW_MEMORY_INTERVAL, DEFAULT_REVIEW_MESSAGE_CHARS, DEFAULT_REVIEW_SKILL_INTERVAL, DEFAULT_REVIEW_TIMEOUT_MS, DEFAULT_SKILL_CONTENT_CHARS, DEFAULT_SKILL_LIMITS, DEFAULT_SKILL_REVIEW_COMPLETION_MIN_TOOL_CALLS, DEFAULT_SKILL_REVIEW_MODEL, DEFAULT_SKILL_REVIEW_TRIGGER, DEFAULT_SUBSTANTIVE_MIN_AGENT_CHARS, DEFAULT_SUBSTANTIVE_MIN_TOOL_CALLS, DEFAULT_SUBSTANTIVE_MIN_USER_CHARS, DEFAULT_USER_CHAR_LIMIT, MAX_TIMER_DELAY_MS, PROMPT_BUNDLE, advanceReview, assertSkillsRootAliasRetired, clampedNumber, clearReviewChannel, contentHash, evolutionIoAdapter, foldToolDispatches, foldTurn, markReviewChannel, newSkillLibrary, readDispatchSignal, redactSecrets, resolveOrigins, resolveRootConfig, reviewPrompt, skillReadNameOf, sweepReviewChannelSessions, verifyPromptBundle } from "@lmzhen/dsh-evolution-core";
 import { validateEvolutionPlan } from "@lmzhen/dsh-evolution-plan-validator";
 //#region lib/types/index.js
 /**
@@ -12,7 +13,7 @@ const name = "evolution-review";
 const inject = ["agents"];
 const Config = z.object({
 	reviewEnabled: z.boolean().default(true),
-	reviewMode: z.union([z.const("subagent"), z.const("inject")]).default("subagent"),
+	reviewMode: z.union([z.const("subagent"), z.const("inject")]).default("inject"),
 	memoryInterval: z.number().min(1).default(DEFAULT_REVIEW_MEMORY_INTERVAL),
 	skillInterval: z.number().min(1).default(DEFAULT_REVIEW_SKILL_INTERVAL),
 	reviewToolAllow: z.array(z.string()).default(["skill"]),
@@ -92,6 +93,25 @@ function apply(ctx, rawConfig = {}) {
 	const cadenceResetWarned = /* @__PURE__ */ new Set();
 	let reviewInFlight = false;
 	const policy = () => ctx.get("evolutionPolicy")?.get();
+	const schemaDefaults = Config["~standard"].validate({}).value;
+	const shadowedRowFields = [
+		"reviewMode",
+		"memoryInterval",
+		"skillInterval"
+	].filter((field) => rawConfig[field] !== void 0 && rawConfig[field] !== schemaDefaults[field]);
+	if (shadowedRowFields.length > 0) {
+		let shadowWarned = false;
+		const warnShadowed = () => {
+			if (shadowWarned) return;
+			shadowWarned = true;
+			ctx.logger.warn(`dsh-evolution-review: this row sets ${shadowedRowFields.join(", ")}, but the mounted evolution-policy service overrides all three — these row values have no effect. Set them on the evolution-policy row instead.`);
+		};
+		if (policy() !== void 0) warnShadowed();
+		else ctx.inject(["evolutionPolicy"], () => {
+			warnShadowed();
+		});
+	}
+	if ((policy()?.reviewMode ?? config.reviewMode) === "inject") ctx.logger.warn("dsh-evolution-review: reviewMode \"inject\" (default) runs the review in the parent session and emits NO evolution/plan-applied ledger entry — evolution-activity and evolution-replay stay empty in this mode (see packages/docs/known-limitations.md). Set reviewMode: \"subagent\" on the evolution-policy row to keep the audited plan path.");
 	const reviewStateLocks = /* @__PURE__ */ new Map();
 	async function withReviewStateLock(id, task) {
 		const next = (reviewStateLocks.get(id) ?? Promise.resolve()).catch(() => {}).then(task);
@@ -104,6 +124,10 @@ function apply(ctx, rawConfig = {}) {
 	}
 	ctx.on("session/event", (session, event) => {
 		if (event.type === "turn/start" && session.header.origin !== "subagent") turnStarts.set(session.id, session.seq - 1);
+		if (event.type === "user/message") {
+			if ((event.data?.source)?.kind === "user") clearReviewChannel(session.id);
+			return;
+		}
 		if (event.type !== "turn/end") return;
 		if (turnStarts.size >= COUNTER_SWEEP_THRESHOLD || cumulativeToolCalls.size >= COUNTER_SWEEP_THRESHOLD || completionInjected.size >= COUNTER_SWEEP_THRESHOLD || pendingCadenceReviews.size >= COUNTER_SWEEP_THRESHOLD || skipNextCadenceFire.size >= COUNTER_SWEEP_THRESHOLD || cadenceResetWarned.size >= COUNTER_SWEEP_THRESHOLD) {
 			const isAlive = (id) => ctx.agents.get(id) !== void 0;
@@ -114,6 +138,7 @@ function apply(ctx, rawConfig = {}) {
 			sweepDeadSessionEntries(pendingCadenceWarned, isAlive);
 			sweepDeadSessionEntries(skipNextCadenceFire, isAlive);
 			sweepDeadSessionEntries(cadenceResetWarned, isAlive);
+			sweepReviewChannelSessions((id) => ctx.agents.get(SessionId(id)) !== void 0);
 		}
 		onTurnEnd(session, event);
 	});
@@ -174,7 +199,7 @@ function apply(ctx, rawConfig = {}) {
 			if (pendingKind !== void 0) {
 				pendingCadenceReviews.delete(session.id);
 				if ((policy()?.reviewMode ?? config.reviewMode) === "inject") {
-					if (!deliverMessage(agent, reviewPrompt(pendingKind), "auto-review")) {
+					if (!deliverMessage(agent, reviewPrompt(pendingKind), "auto-review", true)) {
 						pendingCadenceReviews.set(session.id, pendingKind);
 						return;
 					}
@@ -208,7 +233,7 @@ function apply(ctx, rawConfig = {}) {
 						pendingCadenceReviews.set(session.id, pendingKind);
 						return;
 					} else if (reviewOutcome !== "deferred") {
-						if (!deliverMessage(agent, reviewPrompt(pendingKind), "auto-review")) {
+						if (!deliverMessage(agent, reviewPrompt(pendingKind), "auto-review", true)) {
 							pendingCadenceReviews.set(session.id, pendingKind);
 							return;
 						}
@@ -272,7 +297,7 @@ function apply(ctx, rawConfig = {}) {
 			}
 			return;
 		}
-		if (!deliverMessage(agent, COMPLETION_SKILL_REVIEW_PROMPT, "completion review")) {
+		if (!deliverMessage(agent, COMPLETION_SKILL_REVIEW_PROMPT, "completion review", true)) {
 			completionInjected.delete(session.id);
 			return;
 		}
@@ -302,7 +327,7 @@ function apply(ctx, rawConfig = {}) {
 	* prototype method), the caller's catch demoted that to a console warning,
 	* and the cadence reset ran anyway — the segment's review was consumed with
 	* nothing queued (silent no-delivery window: 2026-09-07 → 0.3.73). */
-	const deliverMessage = (agent, text, summary) => {
+	const deliverMessage = (agent, text, summary, reviewPrompt = false) => {
 		const message = createUserMessage({
 			content: [{
 				type: "text",
@@ -321,6 +346,7 @@ function apply(ctx, rawConfig = {}) {
 				wake.followup(message);
 				skipNextCadenceFire.set(agent.session.id, true);
 			} else agent.inject(message);
+			if (reviewPrompt) markReviewChannel(agent.session.id);
 			return true;
 		} catch (error) {
 			ctx.logger.warn(`dsh-evolution-review: review delivery failed (${error instanceof Error ? error.message : String(error)}) — nothing was queued; the review is NOT consumed and retries at the next completed boundary`);
@@ -354,7 +380,10 @@ function apply(ctx, rawConfig = {}) {
 		const hashes = /* @__PURE__ */ new Map();
 		const io = ctx.get("evolutionIo");
 		if (!io) return hashes;
-		const library = new SkillLibrary(resolveSkillsRoot({ root: rootConfig.root }), evolutionIoAdapter(() => io.provider()));
+		const library = newSkillLibrary({
+			config: rootConfig,
+			io: evolutionIoAdapter(() => io.provider())
+		});
 		for (const summary of await library.list({ withContent: true }).catch((error) => {
 			ctx.logger.warn(`dsh-evolution-review: skill tree scan failed (${error instanceof Error ? error.message : String(error)}) — pre-run staleness hashes are unavailable; full-content updates will not be drift-checked this review`);
 			return [];
@@ -388,6 +417,13 @@ function apply(ctx, rawConfig = {}) {
 			const reviewText = redactSecrets(buildReviewRequest(session, kind, signal, config.reviewContextMessages, config.reviewMessageChars));
 			const agentOptions = { model };
 			if (config.reviewProvider) agentOptions.provider = config.reviewProvider;
+			const registry = ctx.get("tools");
+			const requestedReviewTools = config.reviewToolAllow ?? [];
+			if (requestedReviewTools.length > 0) if (registry === void 0) ctx.logger.warn("dsh-evolution-review: tools service not mounted — the review subagent tool allow-list was not probed; passing it through unchanged");
+			else {
+				const notInGlobalLayer = requestedReviewTools.filter((name) => registry.get(name) == null);
+				if (notInGlobalLayer.length > 0) ctx.logger.warn(`dsh-evolution-review: reviewToolAllow name(s) ${notInGlobalLayer.map((name) => `"${name}"`).join(", ")} are absent from the GLOBAL tool layer — passing them through unchanged (a preset-mounted tool is invisible to a scope-less lookup)`);
+			}
 			const preRunHashes = await treeSkillHashes();
 			const sessionSeqAtPlanTime = session.seq - 1;
 			const run = await subagents.start("spawn", {
@@ -401,7 +437,7 @@ function apply(ctx, rawConfig = {}) {
 				maxDepth: config.reviewMaxDepth,
 				agentOptions,
 				persona: reviewPrompt(kind, "plan"),
-				toolFilter: { allow: [...config.reviewToolAllow ?? []] },
+				toolFilter: { allow: requestedReviewTools },
 				outputSchema: REVIEW_OUTPUT_SCHEMA
 			});
 			try {
@@ -412,9 +448,8 @@ function apply(ctx, rawConfig = {}) {
 					} catch (emitError) {
 						ctx.logger.warn(`dsh-evolution-review: review-error emit failed: ${emitError instanceof Error ? emitError.message : String(emitError)}`);
 					}
-					const failureShape = result;
-					const stopDetail = typeof failureShape.stopReason === "string" ? failureShape.stopReason : void 0;
-					const diagDetail = typeof failureShape.diagnostic === "string" ? failureShape.diagnostic : void 0;
+					const stopDetail = typeof result.stopReason === "string" ? result.stopReason : void 0;
+					const diagDetail = typeof result.diagnostic === "string" ? result.diagnostic : void 0;
 					ctx.logger.warn(`dsh-evolution-review: review subagent returned no structured plan${stopDetail !== void 0 ? ` (stopReason=${stopDetail}${diagDetail !== void 0 ? `; diagnostic=${diagDetail}` : ""})` : ""}`);
 					return false;
 				}
@@ -499,7 +534,7 @@ function apply(ctx, rawConfig = {}) {
 			reviewInFlight = false;
 			const deferred = deferredFallbackReviews.splice(0);
 			for (const { agent: waitingAgent, sessionId: entrySession, kind: waitingKind, prompt, label, channel, counts: entryCounts } of deferred) {
-				if (!deliverMessage(waitingAgent, prompt, label)) {
+				if (!deliverMessage(waitingAgent, prompt, label, true)) {
 					if (channel === "completion") completionInjected.delete(entrySession);
 					continue;
 				}
@@ -525,7 +560,10 @@ function apply(ctx, rawConfig = {}) {
 		const hashLibrary = (() => {
 			const io = ctx.get("evolutionIo");
 			if (!io) return null;
-			return new SkillLibrary(resolveSkillsRoot({ root: rootConfig.root }), evolutionIoAdapter(() => io.provider()));
+			return newSkillLibrary({
+				config: rootConfig,
+				io: evolutionIoAdapter(() => io.provider())
+			});
 		})();
 		const origins = resolveOrigins(void 0, true);
 		const actions = [];
@@ -563,19 +601,17 @@ function apply(ctx, rawConfig = {}) {
 					...op,
 					evidence: op.evidence
 				};
-				let stageCurrent = null;
-				if ((args.action === "update" || args.action === "edit") && hashLibrary && typeof args.name === "string" && args.name !== "") {
-					stageCurrent = await hashLibrary.read(args.name).catch(() => null);
-					if (stageCurrent !== null) args.staged_from_sha256 = contentHash(stageCurrent);
-				}
+				let stagedBytes;
+				if ((args.action === "update" || args.action === "edit") && hashLibrary && typeof args.name === "string" && args.name !== "") stagedBytes = await hashLibrary.read(args.name).catch(() => void 0);
 				if ((args.action === "write_file" || args.action === "remove_file") && hashLibrary && typeof args.name === "string" && args.name !== "" && typeof args.file_path === "string" && args.file_path !== "") {
 					const stageFile = await hashLibrary.readSupportFile(args.name, args.file_path).catch(() => void 0);
 					if (stageFile !== void 0) args.staged_from_sha256 = stageFile === null ? "absent" : contentHash(stageFile);
 				}
 				const opName = typeof args.name === "string" ? args.name : "";
 				const hashChecked = (args.action === "update" || args.action === "edit" || args.action === "patch") && opName !== "" && preRunHashes?.has(opName) === true;
+				const anchored = hashChecked && typeof stagedBytes === "string" ? stagedBytes : null;
 				if (hashChecked) {
-					const live = args.action === "patch" ? hashLibrary ? await hashLibrary.read(opName).catch(() => null) : null : stageCurrent;
+					const live = anchored !== null ? anchored : hashLibrary ? await hashLibrary.read(opName).catch(() => null) : null;
 					const preRun = preRunHashes.get(opName);
 					if (live === null || preRun !== void 0 && contentHash(live) !== preRun) {
 						ok = false;
@@ -583,6 +619,8 @@ function apply(ctx, rawConfig = {}) {
 						continue;
 					}
 				}
+				const anchorBytes = anchored ?? stagedBytes;
+				if ((args.action === "update" || args.action === "edit") && anchorBytes !== void 0) args.staged_from_sha256 = anchorBytes === null ? "absent" : contentHash(anchorBytes);
 				const runnerArgs = {
 					operation: args,
 					origin: origins.library
@@ -670,11 +708,14 @@ function apply(ctx, rawConfig = {}) {
 				message: "evolution-io service not mounted"
 			};
 			const policySnapshot = policySnapshotOf(ctx.get("evolutionPolicy"));
-			const library = new SkillLibrary(resolveSkillsRoot({ root: rootConfig.root }), evolutionIoAdapter(() => io.provider()), {
-				...DEFAULT_SKILL_LIMITS,
-				maxSkillContentChars: policySnapshot?.skillContentChars ?? DEFAULT_SKILL_LIMITS.maxSkillContentChars
-			}, (event) => {
-				ctx.emit("evolution/skill-mutated", event);
+			const library = newSkillLibrary({
+				config: rootConfig,
+				io: evolutionIoAdapter(() => io.provider()),
+				limits: {
+					...DEFAULT_SKILL_LIMITS,
+					maxSkillContentChars: policySnapshot?.skillContentChars ?? DEFAULT_SKILL_LIMITS.maxSkillContentChars
+				},
+				ctx
 			});
 			const op = skillArgs;
 			const name = op.name ?? "";
@@ -792,32 +833,27 @@ function staleRefusal(result, name, filePath) {
 		message: filePath === void 0 ? `Skill "${name}" changed since this plan was produced — the full-content update was refused as stale. Re-read the skill and produce a fresh plan.` : `Support file "${filePath}" of "${name}" changed since this plan was produced — the staged file operation was refused as stale. Re-read the skill tree and produce a fresh plan.`
 	};
 }
+/**
+* v37 P7a: the read-before-write credit now comes from evolution-core's
+* `tool-dispatch` module — the ONE reader of the platform's dispatch event
+* types, and the ONE authority on which tool reads a skill.
+*
+* v32 REV-06(a) is preserved by the normalizer: a skill counts as READ only
+* when it did not fail, so a failed/timeout read still cannot pass the
+* read-before-write gate and let the review blind-overwrite content the model
+* never saw. What changed is the vocabulary the gate listens to: matching
+* `tool/call` here meant every PTC session (`tool/ptc-dispatch*`) collected an
+* EMPTY set, so `filterUnreadSkillOps` dropped every mutating op the model had
+* legitimately read first — and nothing reported the loss.
+* @param session - the session whose log is folded.
+* @returns the skill names this session read through a non-failed dispatch.
+*/
 function collectReadSkillNames(session) {
-	const callNames = /* @__PURE__ */ new Map();
-	const okCallIds = /* @__PURE__ */ new Set();
-	for (const event of session.snapshotEvents()) if (event.type === "tool/call") {
-		if (event.data.name !== "skill") continue;
-		const raw = event.data.arguments;
-		let parsed = {};
-		if (typeof raw === "string") try {
-			parsed = JSON.parse(raw);
-		} catch {
-			continue;
-		}
-		else parsed = raw;
-		const parsedObj = typeof parsed === "object" && parsed !== null ? parsed : {};
-		const name = typeof parsedObj.name === "string" ? parsedObj.name : typeof parsedObj.skill === "string" ? parsedObj.skill : "";
-		if (name) callNames.set(event.data.callId, name);
-	} else if (event.type === "tool/result") {
-		const blocks = event.data?.message?.content;
-		if (!Array.isArray(blocks)) continue;
-		for (const block of blocks) {
-			const typed = block;
-			if (typed.type === "tool-result" && typed.isError !== true && typeof typed.toolCallId === "string") okCallIds.add(typed.toolCallId);
-		}
-	}
 	const names = /* @__PURE__ */ new Set();
-	for (const [callId, name] of callNames) if (okCallIds.has(callId)) names.add(name);
+	for (const dispatch of foldToolDispatches(session.snapshotEvents())) {
+		const name = skillReadNameOf(dispatch);
+		if (name !== void 0) names.add(name);
+	}
 	return names;
 }
 /** Map/set size that triggers a dead-session counter sweep (bounded, not a hard cap). */
@@ -874,7 +910,26 @@ function fingerprintPolicy(snapshot) {
 	}
 }
 /**
-* V10-10 (P2-11): render one `[result]` evidence line from a 'tool/result'
+* The call one log event ANSWERS, in either platform vocabulary.
+*
+* A native tool-result event names its call inside the message (`message.source`,
+* the shape `createToolResultMessage` produces — llm/src/message.ts:258), and a
+* PTC settle event carries `isError` plus the outcome `content` on its own
+* payload. Classified structurally, from the payload, so this file never
+* discriminates on a dispatch event type (arch guard N11).
+* @param event - one persisted session event.
+* @returns the answered call id, or `null` when the event carries no outcome.
+*/
+function resultCallIdOf(event) {
+	const data = event?.data;
+	if (data === void 0 || data === null) return null;
+	const native = data.message?.source?.callId;
+	if (typeof native === "string") return native;
+	if (typeof data.subCallId === "string" && typeof data.isError === "boolean") return data.subCallId;
+	return null;
+}
+/**
+* V10-10 (P2-11): render one `[result]` evidence line from a tool-result
 * event payload. The former read (`data.output`) targeted a field that does
 * not exist on the upstream rc.2 payload, so EVERY result line rendered an
 * empty payload and the review subagent never saw tool output — the evidence
@@ -902,13 +957,21 @@ function buildReviewRequest(session, kind, signal, maxMessages, maxMessageChars)
 	}
 	const toolLines = [];
 	const events = session.snapshotEvents();
+	const openedCallIds = /* @__PURE__ */ new Set();
 	for (let index = events.length - 1; index >= 0 && toolLines.length < 12; index -= 1) {
 		const event = events[index];
-		if (event?.type === "tool/call") {
-			const data = event.data;
-			const argsRaw = typeof data?.arguments === "string" ? data.arguments : JSON.stringify(data?.arguments ?? {});
-			toolLines.push(`[call] ${data?.name ?? "?"} ${argsRaw.slice(0, 500)}`);
-		} else if (event?.type === "tool/result") toolLines.push(renderToolResultLine(event.data));
+		const answeredCallId = resultCallIdOf(event);
+		if (answeredCallId !== null) {
+			if (openedCallIds.has(answeredCallId)) continue;
+			openedCallIds.add(answeredCallId);
+			toolLines.push(renderToolResultLine(event?.data));
+			continue;
+		}
+		const opened = readDispatchSignal(event);
+		if (opened === null || openedCallIds.has(opened.callId)) continue;
+		openedCallIds.add(opened.callId);
+		const argsRaw = typeof opened.arguments === "string" ? opened.arguments : JSON.stringify(opened.arguments ?? {});
+		toolLines.push(`[call] ${opened.name} ${argsRaw.slice(0, 500)}`);
 	}
 	toolLines.reverse();
 	return [

package/lib/types/index.d.ts

@@ -8,7 +8,21 @@ export declare const name = "evolution-review";
 export declare const inject: string[];
 export interface Config {
     reviewEnabled?: boolean;
+    /** How the flush delivers the review: `'inject'` (default since 0.3.74) hands
+     * the review prompt to the PARENT agent — waking it through the same
+     * followup-first channel — so the review runs on the parent's model against
+     * the parent's already-cached prefix and no separate prefill is paid.
+     * `'subagent'` spawns a one-shot child instead: the parent context stays clean
+     * and the review can use a dedicated model, but the child re-prefills its own
+     * system prompt plus a re-serialized conversation digest
+     * (`buildReviewRequest`, redacted, header-prepended) on a DIFFERENT model —
+     * so no prefix cache is shared with the parent and the input tokens are paid
+     * at full price. Deployments that prefer the clean-context split switch the
+     * POLICY row (`evolution-policy`): the mounted policy snapshot shadows this
+     * row's value, which the plugin reports once at load (v37 P2-24). */
     reviewMode?: 'subagent' | 'inject';
+    /** Shadowed by the policy snapshot in every shipped composition — configure
+     * `reviewMemoryInterval` on the `evolution-policy` row instead (v37 P2-24). */
     memoryInterval?: number;
     skillInterval?: number;
     /**
@@ -127,7 +141,7 @@ export declare function filterUnreadSkillOps(ops: Array<{
     name?: string;
 }>, readNames: ReadonlySet<string>): number;
 /**
- * V10-10 (P2-11): render one `[result]` evidence line from a 'tool/result'
+ * V10-10 (P2-11): render one `[result]` evidence line from a tool-result
  * event payload. The former read (`data.output`) targeted a field that does
  * not exist on the upstream rc.2 payload, so EVERY result line rendered an
  * empty payload and the review subagent never saw tool output — the evidence

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@lmzhen/dsh-evolution-review",
   "description": "Background review orchestration (community build)",
-  "version": "0.3.73",
+  "version": "0.3.76",
   "publishConfig": {
     "access": "public"
   },
@@ -18,10 +18,6 @@
       "types": "./lib/types/index.d.ts",
       "default": "./lib/index.js"
     },
-    "./invariant": {
-      "types": "./lib/types/invariant.d.ts",
-      "default": "./lib/invariant.js"
-    },
     "./package.json": "./package.json"
   },
   "files": [
@@ -31,33 +27,31 @@
   "license": "MIT",
   "dependencies": {
     "@deepseek-ai/schemastery": "^3.18.1",
-    "@lmzhen/dsh-evolution-approval": "^0.3.73",
-    "@lmzhen/dsh-evolution-core": "^0.3.73",
-    "@lmzhen/dsh-evolution-plan-validator": "^0.3.73"
+    "@lmzhen/dsh-evolution-approval": "^0.3.76",
+    "@lmzhen/dsh-evolution-core": "^0.3.76",
+    "@lmzhen/dsh-evolution-plan-validator": "^0.3.76"
   },
   "peerDependencies": {
     "@deepseek-ai/cordis": "^4.0.1",
     "@deepseek-ai/dsh-agent": "^0.1.5-rc.2",
-    "@deepseek-ai/dsh-invariants": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-llm": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-session": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-tools": "^0.1.5-rc.2",
-    "@lmzhen/dsh-evolution-state": "^0.3.73",
-    "@lmzhen/dsh-evolution-policy": "^0.3.73"
+    "@lmzhen/dsh-evolution-state": "^0.3.76",
+    "@lmzhen/dsh-evolution-policy": "^0.3.76"
   },
   "devDependencies": {
     "@deepseek-ai/dsh-agent": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-agent-loop-testkit": "^0.1.5-rc.2",
-    "@deepseek-ai/dsh-invariants": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-llm": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-session": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-session-persistence": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-session-persistence-jsonl": "^0.1.5-rc.2",
     "@deepseek-ai/dsh-tools": "^0.1.5-rc.2",
-    "@lmzhen/dsh-evolution-approval": "^0.3.73",
-    "@lmzhen/dsh-evolution-core": "^0.3.73",
-    "@lmzhen/dsh-evolution-curator": "^0.3.73",
-    "@lmzhen/dsh-evolution-plan-validator": "^0.3.73",
-    "@lmzhen/dsh-evolution-state": "^0.3.73"
+    "@lmzhen/dsh-evolution-approval": "^0.3.76",
+    "@lmzhen/dsh-evolution-core": "^0.3.76",
+    "@lmzhen/dsh-evolution-curator": "^0.3.76",
+    "@lmzhen/dsh-evolution-plan-validator": "^0.3.76",
+    "@lmzhen/dsh-evolution-state": "^0.3.76"
   }
 }

package/lib/invariant.js

@@ -1,8 +0,0 @@
-//#region lib/types/invariant.js
-const PACKAGE_NAME = "@lmzhen/dsh-evolution-review";
-const name = "evolution-review-invariant";
-const inject = ["invariants"];
-const install = () => {};
-const apply = (ctx) => Promise.resolve(ctx.invariants.register(PACKAGE_NAME, install));
-//#endregion
-export { apply, inject, name };

package/lib/types/invariant.d.ts

@@ -1,5 +0,0 @@
-import type { Context } from '@deepseek-ai/cordis';
-export declare const name = "evolution-review-invariant";
-export declare const inject: string[];
-export declare const apply: (ctx: Context) => Promise<() => void>;
-//# sourceMappingURL=invariant.d.ts.map