feat(ml): multi-agent context framework + v4 orchestrator prompt

Adds ml/agents/ — five specialised sub-agents (overdue_task, momentum,
time_of_day, recent_patterns, focus_area) each producing a prompt snippet
from user signals. A registry wires them up; the orchestrator prompt in
ml/serving/prompts.py synthesises their outputs into one tip via LiteLLM.

Also wires /api/agents route in the API and updates the Dockerfile to copy
the full ml/ tree with PYTHONPATH=/app so agent imports resolve correctly.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

ml/agents/recent_patterns.py

@@ -0,0 +1,68 @@
+from __future__ import annotations
+from collections import Counter
+from datetime import datetime, timezone
+from typing import ClassVar
+from .base import BaseAgent, AgentInput, AgentOutput
+
+_SEVEN_DAYS_S = 7 * 86_400
+
+
+class RecentPatternsAgent(BaseAgent):
+    """Surfaces the user's reaction pattern from the last 7 days of feedback."""
+    agent_id: ClassVar[str] = "recent-patterns"
+    ttl_seconds: ClassVar[int] = 86_400  # 24h
+    version: ClassVar[str] = "1.0.0"
+
+    def compute(self, inp: AgentInput) -> AgentOutput:
+        now_ts = inp.now.timestamp()
+        recent = [
+            f for f in inp.feedback_history
+            if self._age_s(f.get("created_at", ""), now_ts) <= _SEVEN_DAYS_S
+        ]
+
+        counts: Counter[str] = Counter(f.get("action") for f in recent)
+        total = len(recent)
+        dwell_ms = inp.profile.get("mean_dwell_ms_30d")
+
+        if total == 0:
+            prompt = "No tip reactions recorded in the last 7 days."
+        else:
+            done = counts.get("done", 0)
+            dismissed = counts.get("dismiss", 0)
+            snoozed = counts.get("snooze", 0)
+            parts = [
+                f"Last 7 days: {total} tip reaction{'s' if total != 1 else ''} — "
+                f"{done} completed, {dismissed} dismissed, {snoozed} snoozed."
+            ]
+            if dwell_ms is not None:
+                dwell_s = round(dwell_ms / 1000)
+                if dwell_s < 15:
+                    parts.append(
+                        "Average dwell is very short — user may be acting on auto-pilot; vary tip content."
+                    )
+                elif dwell_s < 60:
+                    parts.append(f"Average dwell {dwell_s}s — tips are being read.")
+                else:
+                    parts.append(
+                        f"Average dwell {dwell_s}s — user deliberates; prefer tips that reward reflection."
+                    )
+            prompt = " ".join(parts)
+
+        snapshot = {
+            "recent_total": total,
+            "action_counts": dict(counts),
+            "mean_dwell_ms_30d": dwell_ms,
+        }
+        return self._make_output(inp, prompt, snapshot)
+
+    @staticmethod
+    def _age_s(iso: str, now_ts: float) -> float:
+        if not iso:
+            return float("inf")
+        try:
+            dt = datetime.fromisoformat(iso.replace("Z", "+00:00"))
+            if dt.tzinfo is None:
+                dt = dt.replace(tzinfo=timezone.utc)
+            return now_ts - dt.timestamp()
+        except Exception:
+            return float("inf")