updates

Author: null-create

.github/agents/Agent-Developer-Assistant/memory/MEMORY.md

@@ -1,5 +1,16 @@
 # Agent Memory — Research Assistant
 
+## 2026-04-22 (Session resume after hard page refresh)
+
+- **Root cause:** `activeSessionIdRef` and `researchConvIdRef` are in-memory `useRef`s — lost on hard refresh. `handleReconnected` was only called on WebSocket re-connections, not on the initial connection after a refresh. Backend fully supports replay but frontend never sent the `resume` message on initial connect.
+- **Fix — frontend only, no backend changes needed:**
+  1. `useWebSocket.ts`: Added `onInitialConnect` callback option (fires on first `ws.onopen`; `onReconnected` fires on subsequent opens). Both stored in refs for stability.
+  2. `App.tsx`: Both `useRef`s now pre-populated via `sessionStorage.getItem()` as initial value — safe because WS `onopen` is async and all React `useEffect`s complete before the network handshake.
+  3. `SS_SESSION_ID` / `SS_CONV_ID` sessionStorage keys written on `session_created` + `handleSendMessage`, cleared on `report`, `research_stopped`, `plan_denied`, `error`.
+  4. `handleInitialConnect`: reads sessionStorage and sends `{ type: 'resume', session_id }` if stored session exists.
+  5. **Replay deduplication:** `replayCountRef = useRef(0)`. Set to `event_count` from `session_resumed`. In message loop, all non-control events decrement counter and set `isReplayedEvent = true` while > 0. `addMessageToConv`, `initFromPlan`, and `addSynthesisNode` guarded with `if (!isReplayedEvent)`. Graph state transitions are NOT guarded — they're idempotent.
+- **Key patterns:** `onInitialConnect` does NOT need the 150ms delay that `onReconnected` has — backend is already up on page load. `session_resumed` reconnect notification message is always added (not guarded) since it's a new message. `error` handler always clears sessionStorage since errors during a resume leave no valid session to return to.
+
 ## 2026-04-15 (Knowledge Graph Schema Redesign — Typed Nodes, Relations, Claims, Documents)
 
 - **Full rewrite of `long_term_memory.py`**: Replaced single `:Entity` node type with 7 typed Neo4j labels (`:Person`, `:Organization`, `:Technology`, `:Concept`, `:Event`, `:Location`, `:Metric`). Each type has its own uniqueness constraint, vector index, and type-specific properties (e.g., Person.affiliation, Technology.version, Metric.value+unit).

backend/api_server.py

@@ -41,7 +41,7 @@
 # ── Config persistence ────────────────────────────────────────────────────────
 _SETTINGS_FILE = Path(
     os.path.join(
-        os.path.dirname(os.path.abspath(__file__)),
+        os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
         "configs",
         "model_settings.json",
     )
@@ -51,7 +51,6 @@
 def _load_persisted_settings() -> dict:
     """Load overrides from the model_settings.json file if it exists."""
     if _SETTINGS_FILE.exists():
-        logger.info("Loading persisted configurations...")
         try:
             return json.loads(_SETTINGS_FILE.read_text())
         except Exception as exc:

backend/config.py

@@ -311,7 +311,7 @@ class Config(BaseModel):
     )
     # Model used for heavy tasks (planning, synthesis, report generation).
     azure_heavy_model: str = Field(
-        default_factory=lambda: os.getenv("AZURE_HEAVY_MODEL", "gpt-5.4")
+        default_factory=lambda: os.getenv("AZURE_HEAVY_MODEL", "gpt-5.2")
     )
     # Model used for light tasks (search execution, analyst extraction).
     azure_light_model: str = Field(
@@ -325,7 +325,7 @@ class Config(BaseModel):
     # Default Bedrock model identifier.
     aws_model: str = Field(
         default_factory=lambda: os.getenv(
-            "AWS_MODEL", "global.anthropic.claude-sonnet-4-6"
+            "AWS_MODEL", "global.anthropic.claude-sonnet-4-5-20250929-v1:0"
         )
     )
     # Model used for heavy tasks (planning, synthesis, report generation).

backend/long_term_memory.py

@@ -4,7 +4,9 @@
 Persistent cross-session memory store backed by Neo4j, with an integrated
 knowledge graph (GraphRAG) for capturing relationships between facts.
 
-Embedding is handled by the shared embeddings.py module (sentence-transformers +
+Replaces the standalone memory MCP server (mcp/memory/) by running the same
+persistence logic in-process alongside the FastAPI backend.  Embedding is
+handled by the shared embeddings.py module (sentence-transformers +
 ThreadPoolExecutor), so there is exactly one model instance and one thread-pool
 in the process.
 
@@ -783,8 +785,11 @@ class KnowledgeGraph:
     Relationship-aware knowledge graph stored natively in Neo4j.
 
     Typed entity nodes (Person, Organization, Technology, Concept, Event,
-    Location, Metric) has its own vector index for semantic search and
-    type-specific properties. Free-form verb phrases are classified via keyword
+    Location, Metric) replace the old generic Entity label.  Each type has
+    its own vector index for semantic search and type-specific properties.
+
+    Typed relationship labels (CAUSES, ENABLES, USES, etc.) replace the old
+    catch-all RELATES_TO.  Free-form verb phrases are classified via keyword
     matching into typed labels; unmatched phrases fall back to RELATES_TO.
 
     Claim nodes store individual assertions linked to entities and documents.

backend/report_exporter.py

@@ -45,11 +45,41 @@ def _get_reports_dir() -> Path:
     return Path(raw)
 
 
-def _sanitize_filename(text: str, max_len: int = 60) -> str:
-    """Turn arbitrary text into a safe filename fragment."""
-    text = re.sub(r"[^\w\s-]", "", text).strip()
-    text = re.sub(r"\s+", "_", text)
-    return text[:max_len] if text else "report"
+# Common English stopwords that add no meaning to a filename.
+_STOPWORDS = frozenset(
+    {
+        "a", "an", "the", "and", "or", "but", "in", "on", "at", "to", "for",
+        "of", "with", "by", "from", "up", "about", "into", "through", "is",
+        "are", "was", "were", "be", "been", "being", "have", "has", "had",
+        "do", "does", "did", "will", "would", "could", "should", "may",
+        "might", "shall", "can", "what", "where", "when", "why", "how",
+        "which", "who", "whom", "this", "that", "these", "those", "it",
+        "its", "we", "they", "you", "i", "my", "our", "your", "their",
+        "his", "her", "some", "any", "all", "each", "as", "if", "than",
+        "so", "yet", "both", "just", "more", "most", "also", "me", "us",
+    }
+)
+
+
+def _query_to_slug(query: str, max_words: int = 5) -> str:
+    """
+    Produce a short, human-readable filename slug from a research query.
+
+    Strips punctuation and common stopwords, takes up to *max_words* of the
+    remaining meaningful terms, title-cases them, and joins with underscores.
+
+    Example: "What are the geopolitical implications of rare earth scarcity?"
+             → "Geopolitical_Implications_Rare_Earth_Scarcity"
+    """
+    # Remove non-alphanumeric characters (keep spaces and hyphens)
+    cleaned = re.sub(r"[^\w\s-]", " ", query).strip()
+    words = cleaned.split()
+    meaningful = [
+        w for w in words if w.lower() not in _STOPWORDS and len(w) > 1
+    ]
+    chosen = meaningful[:max_words] if meaningful else words[:max_words]
+    slug = "_".join(w.capitalize() for w in chosen)
+    return slug if slug else "report"
 
 
 def _extract_title(document: str, query: str) -> str:
@@ -214,8 +244,7 @@ async def export_report_pdf(
     try:
         reports_dir = _get_reports_dir()
         date_str = datetime.now(timezone.utc).strftime("%Y-%m-%d")
-        title = _extract_title(document, query)
-        slug = _sanitize_filename(title)
+        slug = _query_to_slug(query)
         sid_part = (session_id or "cli")[:8]
         filename = f"{date_str}_{sid_part}_{slug}.pdf"
         output_path = reports_dir / filename

backend/requirements.txt

@@ -8,7 +8,6 @@ boto3
 neo4j
 fastapi
 httpx
-fpdf2
 huggingface_hub
 mcp
 ollama