@lmzhen/dsh-evolution-review 0.3.73 → 0.3.74

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.
@@ -38,7 +39,7 @@ 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selects how the flush delivers; the explicit `'inject'` mode's historical "immediate on threshold" contract was superseded in 0.3.39 — both modes are end-of-conversation.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.
 - **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).

package/lib/index.js

@@ -12,7 +12,7 @@ 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"]),

package/lib/types/index.d.ts

@@ -8,6 +8,17 @@ 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 opt back in
+     * per plugin row. */
     reviewMode?: 'subagent' | 'inject';
     memoryInterval?: number;
     skillInterval?: number;

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@lmzhen/dsh-evolution-review",
   "description": "Background review orchestration (community build)",
-  "version": "0.3.73",
+  "version": "0.3.74",
   "publishConfig": {
     "access": "public"
   },
@@ -31,9 +31,9 @@
   "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": "^0.3.74",
+    "@lmzhen/dsh-evolution-core": "^0.3.74",
+    "@lmzhen/dsh-evolution-plan-validator": "^0.3.74"
   },
   "peerDependencies": {
     "@deepseek-ai/cordis": "^4.0.1",
@@ -42,8 +42,8 @@
     "@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": "^0.3.74",
+    "@lmzhen/dsh-evolution-policy": "^0.3.74"
   },
   "devDependencies": {
     "@deepseek-ai/dsh-agent": "^0.1.5-rc.2",
@@ -54,10 +54,10 @@
     "@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": "^0.3.74",
+    "@lmzhen/dsh-evolution-core": "^0.3.74",
+    "@lmzhen/dsh-evolution-curator": "^0.3.74",
+    "@lmzhen/dsh-evolution-plan-validator": "^0.3.74",
+    "@lmzhen/dsh-evolution-state": "^0.3.74"
   }
 }