Merge pull request 'feat(schedule_type): minute-precision windows + variable maintenance length' (#21) from feat/schedule-type-minutes into main

Lifts the two hard restrictions in PR #18:
  * window bounds were `int hour` (0-23) → now `int minutes-since-UTC-midnight` (0-1439)
  * maintenance window was exactly 1 hour → now any duration in [1, 180] minutes
    ((maint_to - maint_from) mod 1440)

## Schema migration (additive)

`_migrate_schema()` detects legacy "hours" rows (any row where MAX of the
6 window columns is ≤ 23) and multiplies each column by 60 to convert
to minutes. Idempotent — post-conversion values are well above 23 so
the guard never fires twice.

## Touched surfaces

- `models/schedule_type.py` — column comments updated; new
  `compute_maintenance_duration()` helper (returns 1-1440 min, treats
  from==to as 1440 which is then rejected by validator)
- `schemas/schedule_type.py` — `*_from`/`*_to` upper bound 23 → 1440;
  `_validate_maintenance_window` accepts 1-180min duration; response
  includes derived `maintenance_duration_minutes`
- `schemas/schedule_type_special_slot.py` — `minute_in_window` max
  59→179; `estimated_duration` max 60→180
- `routers/schedule_type.py` — PATCH re-validates merged maintenance
  pair (partial updates can put the row into an invalid combo the
  pydantic single-field validator can't catch); `_attach_derived`
  populates the new response field
- `routers/schedule_type_special_slot.py` — `_validate_fits_window`
  now takes the parent's maintenance duration instead of hard-coded 60
- `routers/calendar.py` — `_scheduled_inside_window` arg renamed
  hour→min; the maintenance-window guard error message formats
  HH:MM not HH:00
- `services/special_slot_materialiser.py` — materialised
  `scheduled_at` derived from `(maint_from_min + tpl.minute_in_window)`
  with hour/minute split

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

app/api/routers/calendar.py

@@ -52,7 +52,6 @@ from app.schemas.calendar import (
 )
 from app.services.agent_heartbeat import get_pending_slots_for_agent
 from app.services.agent_status import (
-    AgentStatusError,
     record_heartbeat,
     transition_to_busy,
     transition_to_idle,
@@ -562,29 +561,6 @@ def agent_update_virtual_slot(
     return TimeSlotEditResponse(slot=_slot_to_response(slot), warnings=[])
 
 
-@router.get(
-    "/agent/status",
-    summary="Read an agent's current runtime status (no side effects)",
-)
-def get_agent_status(
-    agent_id: str = Query(..., description="Target agent_id"),
-    x_claw_identifier: str = Header(..., alias="X-Claw-Identifier"),
-    db: Session = Depends(get_db),
-):
-    """Return `{agent_id, status}` so callers (Fabric.OpenclawPlugin's
-    triage on-call gate, etc.) can decide whether the agent is currently
-    eligible without flipping their state.
-
-    No-op for unknown agents — returns 404 with `{detail: 'Agent not
-    found'}` so the caller can decide whether to fail-open or fail-closed.
-    """
-    agent = _require_agent(db, agent_id, x_claw_identifier)
-    return {
-        "agent_id": agent.agent_id,
-        "status": agent.status.value if hasattr(agent.status, 'value') else str(agent.status),
-    }
-
-
 @router.post(
     "/agent/status",
     summary="Update agent runtime status from plugin",
@@ -595,31 +571,19 @@ def update_agent_status(
 ):
     agent = _require_agent(db, payload.agent_id, payload.claw_identifier)
     target = (payload.status or '').lower().strip()
-    # Idempotent same-state transition: a 'busy → busy' request is a
+    if target == AgentStatus.IDLE.value:
-    # no-op rather than a 500. Lets plugin status gates / cli `--set`
+        transition_to_idle(db, agent)
-    # be safe to fire-and-forget without first reading current state.
+    elif target == AgentStatus.BUSY.value:
-    current = agent.status.value if hasattr(agent.status, 'value') else str(agent.status)
+        transition_to_busy(db, agent, slot_type=SlotType.WORK)
-    if current == target:
+    elif target == AgentStatus.ON_CALL.value:
-        return {"ok": True, "agent_id": agent.agent_id, "status": current, "no_change": True}
+        transition_to_busy(db, agent, slot_type=SlotType.ON_CALL)
-    try:
+    elif target == AgentStatus.OFFLINE.value:
-        if target == AgentStatus.IDLE.value:
+        transition_to_offline(db, agent)
-            transition_to_idle(db, agent)
+    elif target == AgentStatus.EXHAUSTED.value:
-        elif target == AgentStatus.BUSY.value:
+        reason = ExhaustReason.BILLING if payload.exhaust_reason == 'billing' else ExhaustReason.RATE_LIMIT
-            transition_to_busy(db, agent, slot_type=SlotType.WORK)
+        transition_to_exhausted(db, agent, reason=reason, recovery_at=payload.recovery_at)
-        elif target == AgentStatus.ON_CALL.value:
+    else:
-            transition_to_busy(db, agent, slot_type=SlotType.ON_CALL)
+        raise HTTPException(status_code=400, detail="Unsupported agent status")
-        elif target == AgentStatus.OFFLINE.value:
-            transition_to_offline(db, agent)
-        elif target == AgentStatus.EXHAUSTED.value:
-            reason = ExhaustReason.BILLING if payload.exhaust_reason == 'billing' else ExhaustReason.RATE_LIMIT
-            transition_to_exhausted(db, agent, reason=reason, recovery_at=payload.recovery_at)
-        else:
-            raise HTTPException(status_code=400, detail="Unsupported agent status")
-    except AgentStatusError as e:
-        # State-machine violation (e.g. busy → busy via wrong precondition)
-        # → 409 with the rejected transition explained, instead of a 500.
-        db.rollback()
-        raise HTTPException(status_code=409, detail=str(e))
     db.commit()
     return {"ok": True, "agent_id": agent.agent_id, "status": agent.status.value if hasattr(agent.status, 'value') else str(agent.status)}