init

thirdeye/backend/agents/query_agent.py

@@ -0,0 +1,68 @@
+"""
+Query Agent — voice-aware signal context formatting for ThirdEye.
+
+Provides _format_signal_for_context() which labels each ChromaDB signal with
+its true origin (voice note, document, meeting, chat) so the LLM can produce
+properly attributed answers like:
+  "Based on what @Raj said in a voice note on Mar 14 (45s), the team decided..."
+"""
+from datetime import datetime
+
+
+VOICE_CITATION_INSTRUCTION = """
+When context includes [VOICE NOTE — @name on Date (Xs)] signals, ALWAYS cite the voice note explicitly.
+Example: "Based on what @Raj said in a voice note on Mar 14 (45s), the team decided to use PostgreSQL."
+Never flatten voice signals into generic "the team discussed" language. Always name the speaker and source.
+"""
+
+
+def _format_signal_for_context(signal: dict) -> str:
+    """
+    Format a ChromaDB signal as a context snippet for the Query Agent LLM.
+    Voice-sourced signals get explicit attribution so the LLM cites them correctly.
+    Accepts both flat signal dicts and dicts with a nested 'metadata' key.
+    """
+    # Support both flat dicts and ChromaDB-style {"metadata": {...}, "document": ...}
+    meta = signal.get("metadata", signal)
+
+    source = meta.get("source", signal.get("source", "chat"))
+    sig_type = meta.get("type", signal.get("type", "unknown"))
+    summary = meta.get("summary", signal.get("summary", ""))
+    timestamp = meta.get("timestamp", signal.get("timestamp", ""))
+
+    date_str = ""
+    if timestamp:
+        try:
+            dt = datetime.fromisoformat(timestamp.replace("Z", "+00:00"))
+            date_str = dt.strftime("%b %d")
+        except Exception:
+            date_str = timestamp[:10]
+
+    if source == "voice":
+        speaker = meta.get("speaker", signal.get("speaker", "Unknown"))
+        duration = meta.get("voice_duration", signal.get("voice_duration", 0))
+        duration_str = f"{duration}s" if duration else "?"
+        return (
+            f"[VOICE NOTE — @{speaker} on {date_str} ({duration_str})] "
+            f"[{sig_type}] {summary}"
+        )
+
+    if source == "document":
+        return f"[DOCUMENT — {date_str}] [{sig_type}] {summary}"
+
+    if source == "link":
+        return f"[WEB LINK — {date_str}] [{sig_type}] {summary}"
+
+    if sig_type in ("meet_decision", "meet_action_item", "meet_blocker", "meet_summary"):
+        meeting_id = meta.get("meeting_id", signal.get("meeting_id", ""))
+        return f"[MEETING {meeting_id} — {date_str}] [{sig_type}] {summary}"
+
+    entities_raw = meta.get("entities", signal.get("entities", []))
+    if isinstance(entities_raw, str):
+        import json
+        try:
+            entities_raw = json.loads(entities_raw)
+        except Exception:
+            entities_raw = []
+    sender_str = entities_raw[0] if entities_raw else ""
+    return f"[CHAT — {sender_str} on {date_str}] [{sig_type}] {summary}"