@lmzhen/dsh-evolution-review 0.3.58 → 0.3.59

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (3) hide show
  1. package/README.md +1 -1
  2. package/lib/index.js +1 -12
  3. package/package.json +10 -10
package/README.md CHANGED
@@ -38,6 +38,6 @@ Independent of request-prefix construction. This package does not alter the asse
38
38
  ### Review delivery contract (0.3.38-0.3.42)
39
39
 
40
40
  - **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selects how the flush delivers; the explicit `'inject'` mode's historical "immediate on threshold" contract was superseded in 0.3.39 — both modes are end-of-conversation.
41
- - **`skillReviewTrigger`** (default `'cadence'`): `'cadence'` runs one end-of-conversation review from the cadence latch; `'completion'` uses the separate long-session completion gate (session-cumulative tool-call threshold); `'both'` enables both (note: with the deferred cadence execution the `'both'` combination would inject a second task-complete prompt at the same boundary — prefer `'cadence'`).
41
+ - **`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).
42
42
  - **`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 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.
43
43
  - **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 CHANGED
@@ -192,18 +192,7 @@ function apply(ctx, rawConfig) {
192
192
  if (!shouldCompletionReview(event.data.reason, cumulative, config.skillReviewCompletionMinToolCalls)) return;
193
193
  completionInjected.add(session.id);
194
194
  try {
195
- agent.inject(createUserMessage({
196
- content: [{
197
- type: "text",
198
- text: COMPLETION_SKILL_REVIEW_PROMPT
199
- }],
200
- source: {
201
- kind: "plugin",
202
- plugin: "dsh-evolution-review",
203
- form: "notice",
204
- summary: "completion review"
205
- }
206
- }));
195
+ deliverMessage(agent, COMPLETION_SKILL_REVIEW_PROMPT, "completion review");
207
196
  } catch (injectError) {
208
197
  completionInjected.delete(session.id);
209
198
  ctx.logger.warn(`dsh-evolution-review: completion review inject failed: ${injectError instanceof Error ? injectError.message : String(injectError)}`);
package/package.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "@lmzhen/dsh-evolution-review",
3
3
  "description": "Background review orchestration (community build)",
4
- "version": "0.3.58",
4
+ "version": "0.3.59",
5
5
  "publishConfig": {
6
6
  "access": "public"
7
7
  },
@@ -31,9 +31,9 @@
31
31
  "license": "MIT",
32
32
  "dependencies": {
33
33
  "@deepseek-ai/schemastery": "^3.18.1",
34
- "@lmzhen/dsh-evolution-approval": "^0.3.58",
35
- "@lmzhen/dsh-evolution-core": "^0.3.58",
36
- "@lmzhen/dsh-evolution-plan-validator": "^0.3.58"
34
+ "@lmzhen/dsh-evolution-approval": "^0.3.59",
35
+ "@lmzhen/dsh-evolution-core": "^0.3.59",
36
+ "@lmzhen/dsh-evolution-plan-validator": "^0.3.59"
37
37
  },
38
38
  "peerDependencies": {
39
39
  "@deepseek-ai/cordis": "^4.0.1",
@@ -42,8 +42,8 @@
42
42
  "@deepseek-ai/dsh-llm": "^0.1.1-rc.2",
43
43
  "@deepseek-ai/dsh-session": "^0.1.1-rc.2",
44
44
  "@deepseek-ai/dsh-tools": "^0.1.1-rc.2",
45
- "@lmzhen/dsh-evolution-state": "^0.3.58",
46
- "@lmzhen/dsh-evolution-policy": "^0.3.58"
45
+ "@lmzhen/dsh-evolution-state": "^0.3.59",
46
+ "@lmzhen/dsh-evolution-policy": "^0.3.59"
47
47
  },
48
48
  "devDependencies": {
49
49
  "@deepseek-ai/dsh-agent": "^0.1.1-rc.2",
@@ -54,9 +54,9 @@
54
54
  "@deepseek-ai/dsh-session-persistence": "^0.1.1-rc.2",
55
55
  "@deepseek-ai/dsh-session-persistence-jsonl": "^0.1.1-rc.2",
56
56
  "@deepseek-ai/dsh-tools": "^0.1.1-rc.2",
57
- "@lmzhen/dsh-evolution-approval": "^0.3.58",
58
- "@lmzhen/dsh-evolution-core": "^0.3.58",
59
- "@lmzhen/dsh-evolution-plan-validator": "^0.3.58",
60
- "@lmzhen/dsh-evolution-state": "^0.3.58"
57
+ "@lmzhen/dsh-evolution-approval": "^0.3.59",
58
+ "@lmzhen/dsh-evolution-core": "^0.3.59",
59
+ "@lmzhen/dsh-evolution-plan-validator": "^0.3.59",
60
+ "@lmzhen/dsh-evolution-state": "^0.3.59"
61
61
  }
62
62
  }