fix(sdk): modernize typing and fix ToolResult missing is_error

CI mypy --strict flagged two ToolResult() call sites in tools.py for
missing the `is_error` named argument, and ruff flagged 13 lint issues
in the new sessions.py / test_sessions.py module added in the previous
commit.

- Pass `is_error=` explicitly at both chat_with_tools entry points,
  using `bool(tool_name) and tool_name not in tool_map` so unknown
  tool names are flagged as errors.
- Replace deprecated `typing.Dict` / `typing.List` with `dict` /`list`
  (PEP 585; SDK already targets Python 3.9+ where these work natively).
- Drop unused `Optional` import and `typing.Dict` import.
- Strip quotes from forward-reference `"Phase"` annotation (UP037).
- Replace en-dash in `phase_share` docstring with hyphen-minus
  (RUF002).
- Organize and prune unused imports in tests/test_sessions.py
  (I001, F401).

Author: factorydroid

src/hawk/sessions.py

@@ -13,7 +13,6 @@
 import time
 from dataclasses import dataclass, field
 from enum import Enum
-from typing import Dict, List, Optional
 
 
 class Phase(str, Enum):
@@ -27,7 +26,7 @@ class Phase(str, Enum):
     UNKNOWN = ""
 
     @classmethod
-    def parse(cls, s: str) -> "Phase":
+    def parse(cls, s: str) -> Phase:
         """Return the Phase matching s, or Phase.UNKNOWN if unrecognised."""
         for member in cls:
             if member.value == s:
@@ -54,7 +53,7 @@ class ContextSnapshot:
     token_count: int
     compressed_at: float = field(default_factory=time.time)  # Unix timestamp
     summary: str = ""
-    anchors: List[str] = field(default_factory=list)
+    anchors: list[str] = field(default_factory=list)
 
 
 @dataclass
@@ -84,7 +83,7 @@ class CostAccumulator:
     """
 
     session_id: str
-    by_phase: Dict[Phase, PhaseUsage] = field(default_factory=dict)
+    by_phase: dict[Phase, PhaseUsage] = field(default_factory=dict)
     total_tokens: int = 0
     total_cost_usd: float = 0.0
 
@@ -107,7 +106,7 @@ def add(
         self.total_cost_usd += cost_usd
 
     def phase_share(self, phase: Phase) -> float:
-        """Return the fraction of total tokens consumed by phase (0.0–1.0)."""
+        """Return the fraction of total tokens consumed by phase (0.0-1.0)."""
         if self.total_tokens == 0:
             return 0.0
         usage = self.by_phase.get(phase)
@@ -116,10 +115,16 @@ def phase_share(self, phase: Phase) -> float:
         return usage.total_tokens / self.total_tokens
 
     def __str__(self) -> str:
-        parts = [f"session={self.session_id} total={self.total_tokens}tok ${self.total_cost_usd:.4f}"]
-        for phase, pu in sorted(self.by_phase.items(), key=lambda kv: kv[1].total_tokens, reverse=True):
+        parts = [
+            f"session={self.session_id} total={self.total_tokens}tok ${self.total_cost_usd:.4f}"
+        ]
+        for phase, pu in sorted(
+            self.by_phase.items(), key=lambda kv: kv[1].total_tokens, reverse=True
+        ):
             share = self.phase_share(phase)
-            parts.append(f"  {phase.value or 'unknown'}: {pu.total_tokens}tok ({share:.1%}) ${pu.cost_usd:.4f}")
+            parts.append(
+                f"  {phase.value or 'unknown'}: {pu.total_tokens}tok ({share:.1%}) ${pu.cost_usd:.4f}"
+            )
         return "\n".join(parts)
 
 
@@ -161,9 +166,9 @@ class ResolutionResult:
     """Complete output of a multi-phase resolution run."""
 
     session_id: str = ""
-    candidates: List[PatchCandidate] = field(default_factory=list)
+    candidates: list[PatchCandidate] = field(default_factory=list)
     validation_passed: bool = False
-    validation_failures: List[str] = field(default_factory=list)
-    phase_metrics: List[PhaseMetrics] = field(default_factory=list)
+    validation_failures: list[str] = field(default_factory=list)
+    phase_metrics: list[PhaseMetrics] = field(default_factory=list)
     total_input_tokens: int = 0
     total_output_tokens: int = 0

src/hawk/tools.py

@@ -147,10 +147,13 @@ def chat_with_tools(
                 result = _execute_tool(tool_map[tool_name], arguments)
             else:
                 result = json.dumps({"error": f"Unknown tool: {tool_name}"})
-            tool_results.append(ToolResult(
-                tool_use_id=tc.get("id", "") if isinstance(tc, dict) else getattr(tc, "id", ""),
-                content=result,
-            ))
+            tool_results.append(
+                ToolResult(
+                    tool_use_id=tc.get("id", "") if isinstance(tc, dict) else getattr(tc, "id", ""),
+                    content=result,
+                    is_error=bool(tool_name) and tool_name not in tool_map,
+                )
+            )
 
         # Send tool results back to continue the conversation.
         response = client.chat(
@@ -215,10 +218,13 @@ async def chat_with_tools_async(
                 result = await _execute_tool_async(tool_map[tool_name], arguments)
             else:
                 result = json.dumps({"error": f"Unknown tool: {tool_name}"})
-            tool_results.append(ToolResult(
-                tool_use_id=tc.get("id", "") if isinstance(tc, dict) else getattr(tc, "id", ""),
-                content=result,
-            ))
+            tool_results.append(
+                ToolResult(
+                    tool_use_id=tc.get("id", "") if isinstance(tc, dict) else getattr(tc, "id", ""),
+                    content=result,
+                    is_error=bool(tool_name) and tool_name not in tool_map,
+                )
+            )
 
         response = await client.chat(
             prompt,

tests/test_sessions.py

@@ -1,8 +1,6 @@
 """Tests for hawk.sessions phase-attributed session state."""
 
-import pytest
-
-from hawk.sessions import CostAccumulator, ContextSnapshot, Phase, PhaseUsage, ToolCallRecord
+from hawk.sessions import ContextSnapshot, CostAccumulator, Phase, ToolCallRecord
 
 
 def test_phase_parse_known():