@niadra/sdk 0.10.5 → 0.11.1

package/CHANGELOG.md

@@ -4,6 +4,35 @@ All notable changes to this package are documented here. The format follows [Kee
 
 ## [Unreleased]
 
+## [0.11.1] - 2026-10-05
+
+### Fixed
+
+- The first read of a conversation, task or object in a client gets `timeouts.contextFirst` (1,000 ms, plus
+  the measured round trip) when `context` is left at its default, and so does the first `agentMemory()` read.
+  The API compiles the pack on that read: one right after a customer's first message took 410 ms on the
+  server, the 300 ms default plus the round trip cut it short, and the agent answered without memory. Later
+  reads of the same key keep the short budget. A `context` you set stays a ceiling.
+
+## [0.11.0] - 2026-10-05
+
+The server gives every item one id, a bare UUID, on every surface. This release follows it and removes the
+forms it no longer sends or accepts.
+
+### Changed
+
+- History rows (`search()`, `timeline()`, `open()`) carry each row's own id as a bare UUID; `kind` says what
+  the row is. `open()` takes the `id` of an `episode` or `object` row as listed, and `OpenedItem.kind` is
+  `"episode" | "object"`.
+- `admin.correct()` and `admin.factHistory()` send the fact id as given: a bare UUID. The SDK no longer
+  strips a `fact:` prefix; the server refuses a prefixed id. `feedback()` and `closes` take bare ids too.
+- The canonical tool definitions name the kind of an openable row instead of an id prefix.
+
+### Removed
+
+- The `merged` open item status (`ItemStatus`), `HistoryItem.merged_into`, and `OpenedItem.status` and
+  `merged_into`: the server derives one id per item, so twins are not created and nothing is merged.
+
 ## [0.10.5] - 2026-10-03
 
 ### Added

package/README.md

@@ -287,7 +287,7 @@ if (first) {
 
 `search()` also reports recurrence: how many times the same kind of issue came back, and how it was last resolved.
 
-`filters.when` takes the period in the customer's own words, in Portuguese, English or Spanish (`"semana passada"`, `"last week"`, `"en marzo"`); the answer says in `window` how the server read it, and lists in `ignored` a filter it could not read. Items whose validity ended (an event recorded with `valid_until`, such as an offer valid until Friday) leave reads unless you pass `show_expired: true`. An opened item carries its `versions`, oldest first.
+`filters.when` takes the period in the customer's own words, in Portuguese, English or Spanish (`"semana passada"`, `"last week"`, `"en marzo"`); the answer says in `window` how the server read it, and lists in `ignored` a filter it could not read. Items whose validity ended (an event recorded with `valid_until`, such as an offer valid until Friday) leave reads unless you pass `show_expired: true`. An opened item carries its `versions`, oldest first. Every row's `id` is a bare UUID and `kind` says what it is; `open()` takes the `id` of an `episode` or `object` row, and feedback and `closes` take the `id` of a fact or an open item as listed.
 
 The handle, the search and the conversation id go in request bodies, never in a URL: a conversation id may be a phone number or an e-mail. `open()` sends `POST /v1/history/open`, and the tool kit adds the bound customer to it, so the server opens only that customer's items.
 

package/dist/ai-sdk.cjs

@@ -174,7 +174,7 @@ function render(block, binding, call, mode = "advisory") {
   if (mode === "apply") {
     const disputed = new Set(conflicts.map((c) => c.id));
     const allowed = /* @__PURE__ */ new Set([
-      ...hard.filter((h) => SAID.has(h.source) && (h.scope === "turn" || h.scope === "session") && !disputed.has(h.id)).map((h) => h.id),
+      ...hard.filter((h) => SAID.has(h.source) && (h.scope === "turn" || h.scope === "session" || h.scope === "object") && !disputed.has(h.id)).map((h) => h.id),
       ...attributes.filter((a) => a.source === "stated" || a.source === "correction").map((a) => a.id)
     ]);
     for (const [param, value] of suggested) {