fix(context): reject tool-output byte ceilings below the notice size

A configured tool_output_max_bytes smaller than the truncation notice
itself can't fit a bounded preview, so a persisted result could exceed the
ceiling. Enforce a config floor (ge=1024) so nonsensical values are
rejected at load time instead of being worked around at runtime.

strix/config/settings.py

@@ -69,7 +69,11 @@ class ContextSettings(BaseSettings):
     summary_max_tokens: int = Field(default=4_096, gt=0, alias="STRIX_CONTEXT_SUMMARY_TOKENS")
     tool_output_max_tokens: int = Field(default=8_000, gt=0, alias="STRIX_TOOL_OUTPUT_MAX_TOKENS")
     tool_output_max_lines: int = Field(default=2_000, gt=0, alias="STRIX_TOOL_OUTPUT_MAX_LINES")
-    tool_output_max_bytes: int = Field(default=50 * 1024, gt=0, alias="STRIX_TOOL_OUTPUT_MAX_BYTES")
+    # Floor comfortably above the truncation-notice size so a preview
+    # (head+tail+notice) always fits within the configured ceiling.
+    tool_output_max_bytes: int = Field(
+        default=50 * 1024, ge=1024, alias="STRIX_TOOL_OUTPUT_MAX_BYTES"
+    )
 
 
 class RuntimeSettings(BaseSettings):

strix/tools/output_store.py

@@ -49,7 +49,8 @@ def bound_text(text: str, *, max_lines: int, max_bytes: int) -> str:
     byte size). The removed middle is replaced with a notice recording how
     many lines and bytes were dropped so the agent knows output was elided.
     ``max_bytes`` bounds the *entire* joined result, notice and separators
-    included.
+    included, and must be large enough to hold the notice itself (guaranteed by
+    the ``tool_output_max_bytes`` config floor).
     """
     lines = text.split("\n")
     total_bytes = _byte_len(text)