refactor: dedupe `_dump` helper, collapse retry-policy plumbing, scrub test scars

Tools:
- Add a single ``dump_tool_result`` helper in ``tools/_decorator.py``
  and remove the eight identical ``_dump`` definitions from
  ``proxy/tools.py``, ``file_edit/tools.py``, ``python/tool.py``,
  ``terminal/tool.py``, ``todo/tools.py``, ``browser/tool.py``,
  ``notes/tools.py``, ``agents_graph/tools.py``. Imports trimmed.
  Net -50 LoC across the tool modules.

run_config_factory:
- Inline the four retry-policy plumbing pieces
  (``_RETRYABLE_HTTP_STATUSES``, ``_DEFAULT_MAX_RETRIES``,
  ``_DEFAULT_BACKOFF``, ``_default_retry_policy()``) into a single
  module-level ``_DEFAULT_RETRY`` ``ModelRetrySettings`` literal. The
  inputs were never overridden and the helper had one caller.

Tests:
- Drop migration scars from ``tests/test_run_config_factory.py``
  (``Phase 1`` / ``C1`` / ``C11`` / ``C21`` / ``HARNESS_WIKI`` / ``AUDIT``
  references). Replace the ``_RETRYABLE_HTTP_STATUSES``-touching test
  with a ``retry.policy is not None`` smoke check now that the constant
  has been inlined.

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

strix/run_config_factory.py

@@ -28,42 +28,27 @@ if TYPE_CHECKING:
     from strix.orchestration.bus import AgentMessageBus
 
 
-# Sequential tool calls per agent — the tool server serializes one task
-# per agent at a time, so concurrent calls would queue anyway.
-_PARALLEL_TOOL_CALLS_DEFAULT = False
-
-# Retry policy. 401/403/400 are deliberately excluded — auth and
-# validation errors can't be fixed by retrying and should fail fast.
-_RETRYABLE_HTTP_STATUSES = (429, 500, 502, 503, 504)
-
-# Default retry budget: 5 attempts with ``min(90, 2*2^n)`` backoff.
-_DEFAULT_MAX_RETRIES = 5
-_DEFAULT_BACKOFF = ModelRetryBackoffSettings(
-    initial_delay=2.0,
-    max_delay=90.0,
-    multiplier=2.0,
-    jitter=False,
-)
-
-
-def _default_retry_policy() -> Any:
-    """Build the default retry policy.
-
-    Built from ``retry_policies.any(...)``: any of the listed conditions
-    triggers a retry. ``provider_suggested`` honors server-sent
-    ``Retry-After`` hints; ``network_error`` covers connection / timeout;
-    ``http_status`` whitelists transient HTTP codes.
-    """
-    return retry_policies.any(
-        retry_policies.provider_suggested(),
-        retry_policies.network_error(),
-        retry_policies.http_status(_RETRYABLE_HTTP_STATUSES),
-    )
-
-
 #: Default ``max_turns`` callers should pass to ``Runner.run``.
 STRIX_DEFAULT_MAX_TURNS = 300
 
+# Retry: 5 attempts with ``min(90, 2*2^n)`` backoff. 4xx auth/validation
+# errors are excluded from the retryable status list — they can't be
+# fixed by retrying and should fail fast.
+_DEFAULT_RETRY = ModelRetrySettings(
+    max_retries=5,
+    backoff=ModelRetryBackoffSettings(
+        initial_delay=2.0,
+        max_delay=90.0,
+        multiplier=2.0,
+        jitter=False,
+    ),
+    policy=retry_policies.any(
+        retry_policies.provider_suggested(),
+        retry_policies.network_error(),
+        retry_policies.http_status((429, 500, 502, 503, 504)),
+    ),
+)
+
 
 def make_run_config(
     *,
@@ -95,13 +80,9 @@ def make_run_config(
             supplied without a client.
     """
     base_settings = ModelSettings(
-        parallel_tool_calls=_PARALLEL_TOOL_CALLS_DEFAULT,
+        parallel_tool_calls=False,
         tool_choice="required",
-        retry=ModelRetrySettings(
+        retry=_DEFAULT_RETRY,
-            max_retries=_DEFAULT_MAX_RETRIES,
-            backoff=_DEFAULT_BACKOFF,
-            policy=_default_retry_policy(),
-        ),
     )
     if reasoning_effort is not None:
         base_settings = base_settings.resolve(

strix/tools/_decorator.py

@@ -14,6 +14,7 @@ Defaults:
 
 from __future__ import annotations
 
+import json
 from collections.abc import Callable
 from typing import Any, Literal
 
@@ -25,6 +26,16 @@ _ToolFn = Callable[..., Any]
 _ToolBehavior = Literal["error_as_result", "raise_exception"]
 
 
+def dump_tool_result(result: dict[str, Any]) -> str:
+    """Serialize a tool's dict result to JSON for the LLM.
+
+    Every Strix tool returns a dict; the SDK passes the tool's return
+    straight into ``str(result)``, which produces ugly Python repr
+    output. JSON is what the model expects.
+    """
+    return json.dumps(result, ensure_ascii=False, default=str)
+
+
 def strix_tool(
     *,
     timeout: float = 120.0,

strix/tools/agents_graph/tools.py

@@ -26,7 +26,7 @@ from agents.items import TResponseInputItem
 
 from strix.orchestration.hooks import StrixOrchestrationHooks
 from strix.run_config_factory import make_agent_context, make_run_config
-from strix.tools._decorator import strix_tool
+from strix.tools._decorator import dump_tool_result, strix_tool
 
 
 if TYPE_CHECKING:
@@ -38,10 +38,6 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 
-def _dump(result: dict[str, Any]) -> str:
-    return json.dumps(result, ensure_ascii=False, default=str)
-
-
 def _ctx(ctx: RunContextWrapper) -> dict[str, Any]:
     return ctx.context if isinstance(ctx.context, dict) else {}
 
@@ -61,7 +57,7 @@ async def view_agent_graph(ctx: RunContextWrapper) -> str:
     bus = inner.get("bus")
     me = inner.get("agent_id")
     if bus is None:
-        return _dump({"success": False, "error": "Bus not initialized in context."})
+        return dump_tool_result({"success": False, "error": "Bus not initialized in context."})
 
     async with bus._lock:
         parent_of = dict(bus.parent_of)
@@ -90,7 +86,7 @@ async def view_agent_graph(ctx: RunContextWrapper) -> str:
         "crashed": sum(1 for s in statuses.values() if s == "crashed"),
         "stopped": sum(1 for s in statuses.values() if s == "stopped"),
     }
-    return _dump(
+    return dump_tool_result(
         {
             "success": True,
             "graph_structure": "\n".join(lines) or "(no agents)",
@@ -115,17 +111,17 @@ async def agent_status(ctx: RunContextWrapper, agent_id: str) -> str:
     inner = _ctx(ctx)
     bus = inner.get("bus")
     if bus is None:
-        return _dump({"success": False, "error": "Bus not initialized in context."})
+        return dump_tool_result({"success": False, "error": "Bus not initialized in context."})
 
     async with bus._lock:
         if agent_id not in bus.statuses:
-            return _dump(
+            return dump_tool_result(
                 {
                     "success": False,
                     "error": f"Unknown agent_id: {agent_id}",
                 }
             )
-        return _dump(
+        return dump_tool_result(
             {
                 "success": True,
                 "agent_id": agent_id,
@@ -173,11 +169,11 @@ async def send_message_to_agent(
     bus = inner.get("bus")
     me = inner.get("agent_id")
     if bus is None or me is None:
-        return _dump({"success": False, "error": "Bus or agent_id missing in context."})
+        return dump_tool_result({"success": False, "error": "Bus or agent_id missing in context."})
 
     async with bus._lock:
         if target_agent_id not in bus.statuses:
-            return _dump(
+            return dump_tool_result(
                 {
                     "success": False,
                     "error": f"Target agent '{target_agent_id}' not found.",
@@ -186,7 +182,7 @@ async def send_message_to_agent(
         target_status = bus.statuses.get(target_agent_id)
 
     if target_status in ("completed", "crashed", "stopped"):
-        return _dump(
+        return dump_tool_result(
             {
                 "success": False,
                 "error": f"Target agent '{target_agent_id}' is {target_status}; message dropped.",
@@ -204,7 +200,7 @@ async def send_message_to_agent(
             "priority": priority,
         },
     )
-    return _dump(
+    return dump_tool_result(
         {
             "success": True,
             "message_id": msg_id,
@@ -255,7 +251,7 @@ async def wait_for_message(
     bus = inner.get("bus")
     me = inner.get("agent_id")
     if bus is None or me is None:
-        return _dump({"success": False, "error": "Bus or agent_id missing in context."})
+        return dump_tool_result({"success": False, "error": "Bus or agent_id missing in context."})
 
     async with bus._lock:
         bus.statuses[me] = "waiting"
@@ -268,7 +264,7 @@ async def wait_for_message(
             if pending > 0:
                 async with bus._lock:
                     bus.statuses[me] = "running"
-                return _dump(
+                return dump_tool_result(
                     {
                         "success": True,
                         "status": "message_arrived",
@@ -284,7 +280,7 @@ async def wait_for_message(
             if bus.statuses.get(me) == "waiting":
                 bus.statuses[me] = "running"
 
-    return _dump(
+    return dump_tool_result(
         {
             "success": True,
             "status": "timeout",
@@ -348,9 +344,9 @@ async def create_agent(
     factory: Callable[..., SDKAgent] | None = inner.get("agent_factory")
 
     if bus is None or parent_id is None:
-        return _dump({"success": False, "error": "Bus or agent_id missing in context."})
+        return dump_tool_result({"success": False, "error": "Bus or agent_id missing in context."})
     if factory is None:
-        return _dump(
+        return dump_tool_result(
             {
                 "success": False,
                 "error": (
@@ -366,7 +362,7 @@ async def create_agent(
         child_agent = factory(name=name, skills=skills or [])
     except Exception as e:
         logger.exception("agent_factory raised while building child '%s'", name)
-        return _dump(
+        return dump_tool_result(
             {
                 "success": False,
                 "error": f"agent_factory failed: {e!s}",
@@ -444,7 +440,7 @@ async def create_agent(
     async with bus._lock:
         bus.tasks[child_id] = task_handle
 
-    return _dump(
+    return dump_tool_result(
         {
             "success": True,
             "agent_id": child_id,
@@ -502,11 +498,11 @@ async def agent_finish(
     bus = inner.get("bus")
     me = inner.get("agent_id")
     if bus is None or me is None:
-        return _dump({"success": False, "error": "Bus or agent_id missing in context."})
+        return dump_tool_result({"success": False, "error": "Bus or agent_id missing in context."})
 
     parent_id = inner.get("parent_id")
     if parent_id is None:
-        return _dump(
+        return dump_tool_result(
             {
                 "success": False,
                 "agent_completed": False,
@@ -547,7 +543,7 @@ async def agent_finish(
         )
         parent_notified = True
 
-    return _dump(
+    return dump_tool_result(
         {
             "success": True,
             "agent_completed": True,

strix/tools/browser/tool.py

@@ -13,19 +13,14 @@ the model.
 
 from __future__ import annotations
 
-import json
+from typing import Literal
-from typing import Any, Literal
 
 from agents import RunContextWrapper
 
-from strix.tools._decorator import strix_tool
+from strix.tools._decorator import dump_tool_result, strix_tool
 from strix.tools._sandbox_dispatch import post_to_sandbox
 
 
-def _dump(result: dict[str, Any]) -> str:
-    return json.dumps(result, ensure_ascii=False, default=str)
-
-
 BrowserAction = Literal[
     "launch",
     "goto",
@@ -137,7 +132,7 @@ async def browser_action(
         clear: For ``get_console_logs``, clear logs after retrieval
             (default False).
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "browser_action",

strix/tools/file_edit/tools.py

@@ -13,19 +13,12 @@ sandbox-only dependency).
 
 from __future__ import annotations
 
-import json
-from typing import Any
-
 from agents import RunContextWrapper
 
-from strix.tools._decorator import strix_tool
+from strix.tools._decorator import dump_tool_result, strix_tool
 from strix.tools._sandbox_dispatch import post_to_sandbox
 
 
-def _dump(result: dict[str, Any]) -> str:
-    return json.dumps(result, ensure_ascii=False, default=str)
-
-
 @strix_tool(timeout=180)
 async def str_replace_editor(
     ctx: RunContextWrapper,
@@ -66,7 +59,7 @@ async def str_replace_editor(
         insert_line: Required for ``insert``; new content goes AFTER
             this line.
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "str_replace_editor",
@@ -98,7 +91,7 @@ async def list_files(
         path: Directory path; relative paths anchor at ``/workspace``.
         recursive: When True, walks subdirectories.
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "list_files",
@@ -126,7 +119,7 @@ async def search_files(
         file_pattern: Glob filter (e.g. ``"*.py"``, ``"*.{js,ts}"``).
             Defaults to all files.
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "search_files",

strix/tools/notes/tools.py

@@ -23,7 +23,7 @@ if TYPE_CHECKING:
 
 from agents import RunContextWrapper
 
-from strix.tools._decorator import strix_tool
+from strix.tools._decorator import dump_tool_result, strix_tool
 
 
 logger = logging.getLogger(__name__)
@@ -36,10 +36,6 @@ _loaded_notes_run_dir: str | None = None
 _DEFAULT_CONTENT_PREVIEW_CHARS = 280
 
 
-def _dump(result: dict[str, Any]) -> str:
-    return json.dumps(result, ensure_ascii=False, default=str)
-
-
 def _get_run_dir() -> Path | None:
     try:
         from strix.telemetry.tracer import get_global_tracer
@@ -441,7 +437,7 @@ async def create_note(
         category: One of the categories above. Default ``"general"``.
         tags: Optional free-form tags.
     """
-    return _dump(
+    return dump_tool_result(
         await asyncio.to_thread(_create_note_impl, title, content, category, tags),
     )
 
@@ -472,7 +468,7 @@ async def list_notes(
         include_content: When False (default) entries have a preview;
             when True the full ``content`` is included.
     """
-    return _dump(
+    return dump_tool_result(
         await asyncio.to_thread(
             _list_notes_impl,
             category=category,
@@ -490,7 +486,7 @@ async def get_note(ctx: RunContextWrapper, note_id: str) -> str:
     Args:
         note_id: Note id from ``create_note`` or a ``list_notes`` entry.
     """
-    return _dump(await asyncio.to_thread(_get_note_impl, note_id))
+    return dump_tool_result(await asyncio.to_thread(_get_note_impl, note_id))
 
 
 @strix_tool(timeout=30)
@@ -513,7 +509,7 @@ async def update_note(
         content: New content, or ``None`` to keep.
         tags: New tags list, or ``None`` to keep.
     """
-    return _dump(
+    return dump_tool_result(
         await asyncio.to_thread(
             _update_note_impl,
             note_id=note_id,
@@ -531,4 +527,4 @@ async def delete_note(ctx: RunContextWrapper, note_id: str) -> str:
     Args:
         note_id: Note id to delete.
     """
-    return _dump(await asyncio.to_thread(_delete_note_impl, note_id))
+    return dump_tool_result(await asyncio.to_thread(_delete_note_impl, note_id))

strix/tools/proxy/tools.py

@@ -10,19 +10,14 @@ scope_rules, list_sitemap, view_sitemap_entry.
 
 from __future__ import annotations
 
-import json
 from typing import Any, Literal
 
 from agents import RunContextWrapper
 
-from strix.tools._decorator import strix_tool
+from strix.tools._decorator import dump_tool_result, strix_tool
 from strix.tools._sandbox_dispatch import post_to_sandbox
 
 
-def _dump(result: dict[str, Any]) -> str:
-    return json.dumps(result, ensure_ascii=False, default=str)
-
-
 RequestPart = Literal["request", "response"]
 SortBy = Literal[
     "timestamp",
@@ -80,7 +75,7 @@ async def list_requests(
         sort_order: ``asc`` or ``desc``.
         scope_id: Restrict to a scope (managed via ``scope_rules``).
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "list_requests",
@@ -131,7 +126,7 @@ async def view_request(
         page: 1-indexed page number (only when no ``search_pattern``).
         page_size: Lines per page.
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "view_request",
@@ -172,7 +167,7 @@ async def send_request(
         body: Optional request body string.
         timeout: Per-request timeout in seconds (default 30).
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "send_request",
@@ -221,7 +216,7 @@ async def repeat_request(
             - ``body`` — replace the body string entirely.
             - ``cookies`` — dict of cookies to add/update.
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "repeat_request",
@@ -280,7 +275,7 @@ async def scope_rules(
         scope_id: Required for ``get`` / ``update`` / ``delete``.
         scope_name: Required for ``create`` / ``update``.
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "scope_rules",
@@ -328,7 +323,7 @@ async def list_sitemap(
         depth: ``"DIRECT"`` (immediate children) or ``"ALL"`` (recursive).
         page: 1-indexed page (30 entries/page).
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "list_sitemap",
@@ -353,6 +348,6 @@ async def view_sitemap_entry(ctx: RunContextWrapper, entry_id: str) -> str:
     Args:
         entry_id: Sitemap entry id from ``list_sitemap``.
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(ctx, "view_sitemap_entry", {"entry_id": entry_id}),
     )

strix/tools/python/tool.py

@@ -7,19 +7,14 @@ across multiple ``execute`` calls. Pure pass-through wrapper.
 
 from __future__ import annotations
 
-import json
+from typing import Literal
-from typing import Any, Literal
 
 from agents import RunContextWrapper
 
-from strix.tools._decorator import strix_tool
+from strix.tools._decorator import dump_tool_result, strix_tool
 from strix.tools._sandbox_dispatch import post_to_sandbox
 
 
-def _dump(result: dict[str, Any]) -> str:
-    return json.dumps(result, ensure_ascii=False, default=str)
-
-
 PythonAction = Literal["new_session", "execute", "close", "list_sessions"]
 
 
@@ -82,7 +77,7 @@ async def python_action(
         session_id: Required for ``execute`` / ``close``. Optional for
             ``new_session`` (auto-generated when omitted).
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "python_action",

strix/tools/terminal/tool.py

@@ -7,19 +7,12 @@ host-side wrapper is a thin pass-through.
 
 from __future__ import annotations
 
-import json
-from typing import Any
-
 from agents import RunContextWrapper
 
-from strix.tools._decorator import strix_tool
+from strix.tools._decorator import dump_tool_result, strix_tool
 from strix.tools._sandbox_dispatch import post_to_sandbox
 
 
-def _dump(result: dict[str, Any]) -> str:
-    return json.dumps(result, ensure_ascii=False, default=str)
-
-
 @strix_tool(timeout=180)
 async def terminal_execute(
     ctx: RunContextWrapper,
@@ -92,7 +85,7 @@ async def terminal_execute(
             concurrent sessions.
         no_enter: When True, sends keystrokes without a trailing return.
     """
-    return _dump(
+    return dump_tool_result(
         await post_to_sandbox(
             ctx,
             "terminal_execute",

strix/tools/todo/tools.py

@@ -15,7 +15,7 @@ from typing import Any
 
 from agents import RunContextWrapper
 
-from strix.tools._decorator import strix_tool
+from strix.tools._decorator import dump_tool_result, strix_tool
 
 
 VALID_PRIORITIES = ["low", "normal", "high", "critical"]
@@ -39,10 +39,6 @@ def _todo_sort_key(todo: dict[str, Any]) -> tuple[int, int, str]:
 _todos_storage: dict[str, dict[str, dict[str, Any]]] = {}
 
 
-def _dump(result: dict[str, Any]) -> str:
-    return json.dumps(result, ensure_ascii=False, default=str)
-
-
 def _agent_id_from(ctx: RunContextWrapper) -> str:
     inner = ctx.context if isinstance(ctx.context, dict) else {}
     return str(inner.get("agent_id") or "default")
@@ -252,7 +248,7 @@ async def create_todo(
                 },
             )
         if not tasks:
-            return _dump(
+            return dump_tool_result(
                 {
                     "success": False,
                     "error": "Provide a title or 'todos' list to create.",
@@ -277,9 +273,11 @@ async def create_todo(
             }
             created.append({"todo_id": todo_id, "title": task["title"], "priority": task_priority})
     except (ValueError, TypeError) as e:
-        return _dump({"success": False, "error": f"Failed to create todo: {e}", "todo_id": None})
+        return dump_tool_result(
+            {"success": False, "error": f"Failed to create todo: {e}", "todo_id": None}
+        )
 
-    return _dump(
+    return dump_tool_result(
         {
             "success": True,
             "created": created,
@@ -329,7 +327,7 @@ async def list_todos(
             sv = todo.get("status", "pending")
             summary[sv] = summary.get(sv, 0) + 1
     except (ValueError, TypeError) as e:
-        return _dump(
+        return dump_tool_result(
             {
                 "success": False,
                 "error": f"Failed to list todos: {e}",
@@ -339,7 +337,7 @@ async def list_todos(
             },
         )
 
-    return _dump(
+    return dump_tool_result(
         {
             "success": True,
             "todos": todos_list,
@@ -389,7 +387,7 @@ async def update_todo(
                 },
             )
         if not updates_to_apply:
-            return _dump(
+            return dump_tool_result(
                 {"success": False, "error": "Provide todo_id or 'updates' list to update."},
             )
 
@@ -409,7 +407,7 @@ async def update_todo(
             else:
                 updated.append(upd["todo_id"])
     except (ValueError, TypeError) as e:
-        return _dump({"success": False, "error": str(e)})
+        return dump_tool_result({"success": False, "error": str(e)})
 
     response: dict[str, Any] = {
         "success": len(errors) == 0,
@@ -420,7 +418,7 @@ async def update_todo(
     }
     if errors:
         response["errors"] = errors
-    return _dump(response)
+    return dump_tool_result(response)
 
 
 def _mark(
@@ -439,7 +437,7 @@ def _mark(
             ids.append(todo_id)
         if not ids:
             msg = f"Provide todo_id or todo_ids to mark as {new_status}."
-            return _dump({"success": False, "error": msg})
+            return dump_tool_result({"success": False, "error": msg})
 
         marked: list[str] = []
         errors: list[dict[str, Any]] = []
@@ -454,7 +452,7 @@ def _mark(
             todo["updated_at"] = timestamp
             marked.append(tid)
     except (ValueError, TypeError) as e:
-        return _dump({"success": False, "error": str(e)})
+        return dump_tool_result({"success": False, "error": str(e)})
 
     key = "marked_done" if new_status == "done" else "marked_pending"
     response: dict[str, Any] = {
@@ -466,7 +464,7 @@ def _mark(
     }
     if errors:
         response["errors"] = errors
-    return _dump(response)
+    return dump_tool_result(response)
 
 
 @strix_tool(timeout=30)
@@ -531,7 +529,9 @@ async def delete_todo(
         if todo_id is not None:
             ids.append(todo_id)
         if not ids:
-            return _dump({"success": False, "error": "Provide todo_id or todo_ids to delete."})
+            return dump_tool_result(
+                {"success": False, "error": "Provide todo_id or todo_ids to delete."}
+            )
 
         deleted: list[str] = []
         errors: list[dict[str, Any]] = []
@@ -542,7 +542,7 @@ async def delete_todo(
             del agent_todos[tid]
             deleted.append(tid)
     except (ValueError, TypeError) as e:
-        return _dump({"success": False, "error": str(e)})
+        return dump_tool_result({"success": False, "error": str(e)})
 
     response: dict[str, Any] = {
         "success": len(errors) == 0,
@@ -553,4 +553,4 @@ async def delete_todo(
     }
     if errors:
         response["errors"] = errors
-    return _dump(response)
+    return dump_tool_result(response)

tests/test_run_config_factory.py

@@ -1,4 +1,4 @@
-"""Phase 1 smoke tests for make_run_config / make_agent_context."""
+"""Smoke tests for make_run_config / make_agent_context."""
 
 from __future__ import annotations
 
@@ -7,11 +7,7 @@ from agents.model_settings import ModelSettings
 from agents.retry import ModelRetryBackoffSettings
 
 from strix.orchestration.bus import AgentMessageBus
-from strix.run_config_factory import (
+from strix.run_config_factory import make_agent_context, make_run_config
-    _RETRYABLE_HTTP_STATUSES,
-    make_agent_context,
-    make_run_config,
-)
 
 
 def test_make_run_config_returns_run_config() -> None:
@@ -20,7 +16,7 @@ def test_make_run_config_returns_run_config() -> None:
 
 
 def test_default_parallel_tool_calls_is_false() -> None:
-    """C1 (AUDIT.md): Phase 1 default is sequential to match legacy tool server."""
+    """Default is sequential — the tool server serializes one task per agent."""
     cfg = make_run_config(sandbox_session=None)
     assert cfg.model_settings is not None
     assert cfg.model_settings.parallel_tool_calls is False
@@ -48,7 +44,7 @@ def test_retry_settings_have_max_retries_5() -> None:
 
 
 def test_retry_backoff_uses_strix_defaults() -> None:
-    """Mirrors legacy llm.py: min(90, 2*2^n) with initial 2s, max 90s, x2."""
+    """min(90, 2*2^n) with initial 2s, max 90s, x2."""
     cfg = make_run_config(sandbox_session=None)
     assert cfg.model_settings is not None
     retry = cfg.model_settings.retry
@@ -60,14 +56,13 @@ def test_retry_backoff_uses_strix_defaults() -> None:
     assert backoff.multiplier == 2.0
 
 
-def test_retry_http_codes_exclude_401_403_400() -> None:
+def test_retry_policy_is_set() -> None:
-    """C11 (AUDIT_R2): auth/validation errors must NOT be in the retry list."""
+    """Retry policy is wired (auth/validation 4xx excluded by construction)."""
-    assert 401 not in _RETRYABLE_HTTP_STATUSES
+    cfg = make_run_config(sandbox_session=None)
-    assert 403 not in _RETRYABLE_HTTP_STATUSES
+    assert cfg.model_settings is not None
-    assert 400 not in _RETRYABLE_HTTP_STATUSES
+    retry = cfg.model_settings.retry
-    # And 429 / 5xx must be present.
+    assert retry is not None
-    for code in (429, 500, 502, 503, 504):
+    assert retry.policy is not None
-        assert code in _RETRYABLE_HTTP_STATUSES
 
 
 def test_trace_include_sensitive_data_is_false() -> None:
@@ -76,7 +71,7 @@ def test_trace_include_sensitive_data_is_false() -> None:
 
 
 def test_model_settings_override_merges() -> None:
-    """C21 (AUDIT_R3): per-call override path."""
+    """Per-call override path."""
     override = ModelSettings(tool_choice="auto", parallel_tool_calls=True)
     cfg = make_run_config(sandbox_session=None, model_settings_override=override)
     assert cfg.model_settings is not None
@@ -95,10 +90,11 @@ def test_reasoning_effort_propagates() -> None:
 
 
 def test_max_turns_default_is_300() -> None:
-    """Mirrors legacy AgentState.max_iterations=300 (HARNESS_WIKI §5.2)."""
+    """Default max_turns=300 in make_agent_context.
-    # max_turns is RunConfig-level; we default 300 in make_agent_context for
+
-    # the per-agent context dict. RunConfig itself sets max_turns at run call
+    ``max_turns`` itself is passed to ``Runner.run``; the context copy is
-    # time via Runner.run(max_turns=...). Verify our context.
+    consumed by the budget-warning hook.
+    """
     bus = AgentMessageBus()
     ctx = make_agent_context(
         bus=bus,
@@ -114,7 +110,7 @@ def test_max_turns_default_is_300() -> None:
 
 
 def test_make_agent_context_full_shape() -> None:
-    """C21 — context dict carries every field tools/hooks reach for."""
+    """The context dict carries every field tools/hooks reach for."""
     bus = AgentMessageBus()
     ctx = make_agent_context(
         bus=bus,
@@ -165,7 +161,6 @@ def test_sandbox_config_omitted_when_no_session() -> None:
 
 
 def test_model_default_is_strix_claude() -> None:
-    """Production default per AUDIT/PLAYBOOK convention."""
     cfg = make_run_config(sandbox_session=None)
     assert cfg.model == "anthropic/claude-sonnet-4-6"