@lmzhen/dsh-evolution-feedback 0.3.83 → 0.4.0

package/README.md

@@ -1,27 +1,27 @@
 # @lmzhen/dsh-evolution-feedback
 
-Feedback-to-quality scoring for self-evolution
+Feedback-to-quality scoring: durable through `ctx.evolutionIo`; skill feedback feeds the
+FEEDBACK-OWNED `feedback_score`/`feedback_warn` pair.
 
-## Model Experience
+## Model surface
 
-### Indirect model surface
+- **Model-visible:** nothing of its own: the usage-record readers own it.
+- **Prompt prefix / KV cache:** unchanged by this package: family-level rules single-sourced in `packages/README.md` §"Model-visible prompt prefix and the KV cache".
+- **Mount it?** yes — the `evolution-feedback` row, `evolution-host`/`evolution-all`/`evolution-preset`.
 
-#### What the model sees
+## Configuration
 
-`@lmzhen/dsh-evolution-feedback` registers no direct prompt or tool schema itself. Model-visible effects are owned by the packages that consume this service.
+- `qualityWarnThreshold` (default `-0.25`): score below which the pair flips to warned.
+- `path`: boot-cache path; `''` means `$DSH_HOME/evolution/feedback.json` (the event log is not affected).
 
-#### Token effect
+## Known limitations
 
-Zero direct token effect from this package; consumers add any model-visible tokens.
-
-#### KV Cache effect
-
-Independent of request-prefix construction. This package does not alter the assembled prompt or tool list.
-
-## Known Limitations and Deferred Work
+- Lifecycle and scope views read `quality_warn || feedback_warn`, so negative feedback shortens the stale window even though the curator recomputes `quality_warn`.
 
+**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).
 
-- Persists through the IO seam; quality propagation into skill usage requires the `skillUsage` service.
-- P1-1 (v15): feedback writes the FEEDBACK-OWNED `feedback_score`/`feedback_warn` usage fields. The lifecycle engine and the scope view read the union `quality_warn || feedback_warn`, so a negative feedback shortens the stale window even though the curator's six-factor run always recomputes `quality_warn` itself.
+## Notes and history
 
-**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).
+- Persists through the IO seam; quality propagation into skill usage requires BOTH the `skillUsage` service and a mounted IO backend: with no `evolutionIo` the push returns before touching the persistent usage sidecar (S4.6), so an optimistic in-memory score never lands there unconfirmed.
+- The in-memory-wins window of a log refold is the IN-FLIGHT append window only, counted PER TARGET (two appends for one target in flight at once both stay protected, and the no-io path never enters it): a settled target folds from the log truth, so another process sharing `DSH_HOME` is not overwritten by this process's stale record (S1.4).
+- P1-1 (v15): feedback writes the FEEDBACK-OWNED `feedback_score`/`feedback_warn` usage fields.

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@lmzhen/dsh-evolution-feedback",
   "description": "Feedback-to-quality scoring for self-evolution (community build)",
-  "version": "0.3.83",
+  "version": "0.4.0",
   "publishConfig": {
     "access": "public"
   },
@@ -27,16 +27,16 @@
   "license": "MIT",
   "dependencies": {
     "@deepseek-ai/schemastery": "^3.18.1",
-    "@lmzhen/dsh-evolution-core": "^0.3.83"
+    "@lmzhen/dsh-evolution-core": "^0.4.0"
   },
   "peerDependencies": {
     "@deepseek-ai/cordis": "^4.0.1",
-    "@lmzhen/dsh-evolution-io": "^0.3.83",
-    "@lmzhen/dsh-skill-usage": "^0.3.83"
+    "@lmzhen/dsh-evolution-io": "^0.4.0",
+    "@lmzhen/dsh-skill-usage": "^0.4.0"
   },
   "devDependencies": {
-    "@lmzhen/dsh-evolution-io": "^0.3.83",
-    "@lmzhen/dsh-evolution-io-node": "^0.3.83",
-    "@lmzhen/dsh-skill-usage": "^0.3.83"
+    "@lmzhen/dsh-evolution-io": "^0.4.0",
+    "@lmzhen/dsh-evolution-io-node": "^0.4.0",
+    "@lmzhen/dsh-skill-usage": "^0.4.0"
   }
 }