nexo-brain 7.9.30 → 7.9.31

package/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "nexo-brain",
-  "version": "7.9.30",
+  "version": "7.9.31",
   "description": "Local cognitive runtime for Claude Code \u2014 persistent memory, overnight learning, doctor diagnostics, personal scripts, recovery-aware jobs, startup preflight, and optional dashboard/power helper.",
   "author": {
     "name": "NEXO Brain",

package/README.md

@@ -18,7 +18,9 @@
 
 [Watch the overview video](https://nexo-brain.com/watch/) · [Watch on YouTube](https://www.youtube.com/watch?v=i2lkGhKyVqI) · [Open the infographic](https://nexo-brain.com/assets/nexo-brain-infographic-v5.png)
 
-Version `7.9.30` is the current packaged-runtime line. Patch release over `7.9.29`: hotfix for a missing ``import sys`` in ``src/agent_runner.py`` that ruff F821 caught in CI and blocked the 7.9.29 publish workflow before any npm artifact shipped. ``nexo-brain@7.9.30`` is the first npm release that carries the 7.9.29 override-path hardening.
+Version `7.9.31` is the current packaged-runtime line. Patch release over `7.9.30`: fixes a wire-level bug where ``call_model_raw`` was sending ``stop_sequences=["\n", ".", " "]`` by default, which the current Anthropic Messages API rejects with HTTP 400 ``each stop sequence must contain non-whitespace``. The default is now ``None`` (no ``stop_sequences`` field sent) since ``max_tokens=3`` already caps the yes/no classifier output. A local guard rejects whitespace-only caller values up front so the error shows where the caller is, not as a remote 400. Also removes an internal design document that did not belong in the open-source distribution.
+
+Previously in `7.9.30`: hotfix for a missing ``import sys`` in ``src/agent_runner.py`` that ruff F821 caught in CI and blocked the 7.9.29 publish workflow before any npm artifact shipped. ``nexo-brain@7.9.30`` is the first npm release that carries the 7.9.29 override-path hardening.
 
 Previously in `7.9.29`: hardening pass on the optional LLM endpoint and auth provider override path. The bearer is now passed to the Anthropic SDK via `auth_token` so it lands in the standard `Authorization: Bearer` header (7.9.28 sent it as `X-Api-Key` and any compatible proxy rejected every request with 401). The Brain config directory is resolved on each call instead of cached at import, so LaunchAgent crons that export `NEXO_HOME` via a wrapper now reach the right `~/.nexo/config/`. The `Idempotency-Key` header accepts a caller-provided value so application-level retries reuse the same dedup key. Override mode is strict about its bearer source: if `auth_provider.json` is missing or the helper fails, the call raises `ClassifierUnavailableError` instead of falling back to the operator's real `ANTHROPIC_API_KEY`, which would otherwise leak to the custom proxy as a second header. A new end-to-end test suite drives the real SDK against a local `http.server` and asserts on captured wire headers and body, complementing the SDK-mock unit tests.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "nexo-brain",
-  "version": "7.9.30",
+  "version": "7.9.31",
   "mcpName": "io.github.wazionapps/nexo",
   "description": "NEXO Brain — Shared brain for AI agents. Persistent memory, semantic RAG, natural forgetting, metacognitive guard, trust scoring, 150+ MCP tools. Works with Claude Code, Codex, Claude Desktop & any MCP client. 100% local, free.",
   "homepage": "https://nexo-brain.com",

package/src/call_model_raw.py

@@ -469,9 +469,16 @@ def _call_anthropic_raw(
         "model": wire_model,
         "max_tokens": max_tokens,
         "temperature": temperature,
-        "stop_sequences": stop_sequences,
         "messages": [{"role": "user", "content": prompt}],
     }
+    if stop_sequences:
+        # Anthropic API rejects whitespace-only stop sequences with
+        # 400 ``each stop sequence must contain non-whitespace``. The
+        # caller-validation in call_model_raw filters these out before
+        # we reach this point; the empty/None case is also covered by
+        # the truthy check above so we omit the field entirely instead
+        # of sending ``stop_sequences: null`` to the wire.
+        kwargs["stop_sequences"] = stop_sequences
     if system:
         kwargs["system"] = system
 
@@ -582,7 +589,20 @@ def call_model_raw(
                           "enforcer_classifier".
         max_tokens      — hard cap on output tokens. Default 3 (yes/no only).
         temperature     — sampling temperature. Default 0.0 (deterministic).
-        stop_sequences  — early-stop strings. Default ["\\n", ".", " "].
+        stop_sequences  — early-stop strings. Default ``None`` (no stop
+                          sequence sent on the wire). Anthropic's API
+                          rejects whitespace-only entries with
+                          ``each stop sequence must contain
+                          non-whitespace`` (HTTP 400), so the previous
+                          default of ``["\\n", ".", " "]`` made every
+                          ``enforcer_classifier`` request fail in
+                          production. ``max_tokens=3`` already serves as
+                          the hard cap for yes/no classification, so a
+                          stop sequence is unnecessary by default.
+                          Callers that want a deterministic stop can
+                          pass e.g. ``["."]``; whitespace-only entries
+                          are rejected locally with
+                          ``ClassifierUnavailableError``.
         timeout         — per-request timeout in seconds. Default 10.0.
         system          — optional system prompt. Default None (provider default).
         idempotency_key — optional opaque token attached as
@@ -612,8 +632,22 @@ def call_model_raw(
     Callers MUST catch this and fall back to a safer default. Fase 2 spec
     0.20 is explicit: silence is not obedience. Never fail-open.
     """
-    if stop_sequences is None:
-        stop_sequences = ["\n", ".", " "]
+    if stop_sequences is not None:
+        # Anthropic API: ``each stop sequence must contain
+        # non-whitespace`` (HTTP 400). Surface the configuration error
+        # locally instead of letting Anthropic 400 the request — and,
+        # in override mode, instead of letting the proxy translate that
+        # 400 into a misleading ``503 all_providers_down``.
+        invalid = [
+            repr(s) for s in stop_sequences
+            if not isinstance(s, str) or not s.strip()
+        ]
+        if invalid:
+            raise ClassifierUnavailableError(
+                "stop_sequences contains whitespace-only or non-string "
+                f"entries: {', '.join(invalid)}; Anthropic API requires "
+                "every stop sequence to contain non-whitespace"
+            )
 
     # Local imports to avoid circulars and keep agent_runner.py decoupled.
     from client_preferences import (  # type: ignore