@tangle-network/agent-knowledge 22.0.18 → 23.0.1

package/docs/agent-tools.md

@@ -32,7 +32,9 @@ It does not report a partial write as tool success.
 The lower-level `applyKnowledgeWriteBlocks` API retains its explicit `written` and `warnings` result for callers that inspect partial proposals.
 `knowledge_resolve` returns the resolution status of each reference.
 Every tool result carries `timing` in milliseconds: `viewMs` to obtain the current page view for search, read and resolve, `visibilityMs` to persist a search's snapshot, and `lockWaitMs` and `lockHoldMs` for a record.
-Tools re-read only the page files whose identity changed since the previous call, and concurrent calls share one refresh.
+Search, read and resolve also count `pagesRead`, the files read from disk to refresh the view, when the stores come from `createRunScopedStores`.
+Every page read in a process, including `loadKnowledgePages` and the write intake, re-reads only the page files whose identity (device, inode, size, mtime, ctime) changed since the previous read, and concurrent calls share one refresh.
+A file changed within the last second is re-read until it settles, because a rewrite within one filesystem timestamp tick can keep its identity.
 
 Knowledge tools retain write history by default.
 Each transaction records the host actorId, runId, and exact prior and new bytes.

package/docs/compatibility.md

@@ -1,15 +1,15 @@
 # Compatibility and peer versions
 
-Knowledge 22 requires Node.js 22.12 or later and Interface 5.
+Knowledge 23 requires Node.js 22.12 or later, Interface 5 and Eval 1.
 
-`package.json` owns the supported Eval versions: the `@tangle-network/agent-eval` peer
-range and its exact development pin. The supported minors run from the range's floor through
-the pin's minor; all depend on Core 0.12, Interface 5 and Trace Contract 2.
-`scripts/lib/peer-range.mjs` derives the list, and packed-package checks run once per minor,
-at x.y.0 and at the pin for the last, each with one physical copy of Eval, Core 0.12,
-Interface 5 and Zod 4.6.5. Official optimizer checks require the matching `agent-eval-rpc`
-Python bridge, `==<version>`, for each.
+`package.json` owns the supported Eval versions: the `@tangle-network/agent-eval` caret peer
+range and its exact development pin. Eval 1 states that a minor is additive, so the range admits
+every later 1.x release. `scripts/lib/peer-range.mjs` derives the checked versions, the range's
+floor and the pin, and packed-package checks run once for each, with one physical copy of Eval,
+Core, Interface and Zod. Official optimizer checks require the matching `agent-eval-rpc` Python
+bridge, `==<version>`, for each.
 
+Use Knowledge 22 with Eval 0.218 through 0.222.
 Use Knowledge 21 with Interface 4 and Eval 0.212. Use Knowledge 20 with Interface 3
 and Eval 0.209 through 0.211. Use Knowledge 19 with Interface 2 and Eval 0.201
 through 0.208, and Knowledge 17 with older Eval releases.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-knowledge",
-  "version": "22.0.18",
+  "version": "23.0.1",
   "description": "Build, search, evaluate, and improve source-backed knowledge bases.",
   "homepage": "https://github.com/tangle-network/agent-knowledge#readme",
   "repository": {
@@ -85,15 +85,15 @@
     "zod": "^4.6.6"
   },
   "peerDependencies": {
-    "@tangle-network/agent-eval": ">=0.221.1 <0.223.0",
+    "@tangle-network/agent-eval": "^1.0.0",
     "@tangle-network/agent-interface": "^5.2.0"
   },
   "devDependencies": {
     "@arethetypeswrong/cli": "^0.18.5",
     "@biomejs/biome": "^2.6.2",
     "@neo4j-labs/agent-memory": "0.6.0",
-    "@tangle-network/agent-eval": "0.222.0",
-    "@tangle-network/agent-interface": "5.8.1",
+    "@tangle-network/agent-eval": "1.0.0",
+    "@tangle-network/agent-interface": "5.10.2",
     "@types/node": "^26.6.5",
     "mem0ai": "3.3.1",
     "oxc-parser": "0.155.0",