v0.5.1 — gate show_reasoning behind opt-in flag

reasoning_content blocks can be large; default to off on advise tool
and CLI (--show-reasoning to enable). Matches original spec intent.

Author: arizen-dev

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "deepseek-mcp"
-version = "0.5.0"
+version = "0.5.1"
 description = "A tiny MCP stdio server that exposes DeepSeek as a Claude/Codex tool."
 readme = "README.md"
 requires-python = ">=3.10"

src/deepseek_mcp/__main__.py

@@ -33,6 +33,10 @@ def main() -> None:
     advise_parser.add_argument(
         "--effort", choices=["medium", "high", "max"], default="max",
     )
+    advise_parser.add_argument(
+        "--show-reasoning", action="store_true",
+        help="Include <reasoning> block from thinking mode in output",
+    )
 
     args = parser.parse_args()
 
@@ -81,7 +85,12 @@ def _cmd_advise(args: argparse.Namespace) -> None:
         sys.exit(2)
 
     from deepseek_mcp.server import call_advisor
-    call_args = {"prompt": args.prompt, "effort": args.effort, "system": None}
+    call_args = {
+        "prompt": args.prompt,
+        "effort": args.effort,
+        "system": None,
+        "show_reasoning": args.show_reasoning,
+    }
     result = call_advisor(call_args)
     print(result)
 

src/deepseek_mcp/server.py

@@ -14,7 +14,7 @@
 
 PROTOCOL_VERSION = "2024-11-05"
 SERVER_NAME = "deepseek-mcp"
-SERVER_VERSION = "0.5.0"
+SERVER_VERSION = "0.5.1"
 
 PRICING: dict[str, dict[str, float]] = {
     "deepseek-v4-flash": {"in": 0.14, "out": 0.28},
@@ -146,6 +146,15 @@ def _log_call(entry: dict[str, Any]) -> None:
                         "medium: lighter thinking, quicker (~30s)."
                     ),
                 },
+                "show_reasoning": {
+                    "type": "boolean",
+                    "default": False,
+                    "description": (
+                        "If true, prepend a <reasoning>...</reasoning> block "
+                        "with the model's chain-of-thought. Off by default to "
+                        "keep responses compact — reasoning_content can be large."
+                    ),
+                },
             },
             "required": ["prompt"],
         },
@@ -242,7 +251,7 @@ def call_advisor(args: dict[str, Any], progress_token: Any = None) -> str:
         "latency_s": round(time.time() - started_at, 2),
         "cost_usd": round(cost, 6) if cost else None,
     })
-    if reasoning_text:
+    if reasoning_text and args.get("show_reasoning"):
         result = f"<reasoning>\n{reasoning_text}\n</reasoning>\n\n{result}"
     return result
 

tests/test_protocol.py

@@ -15,7 +15,7 @@ def test_initialize_response():
     assert response["id"] == 1
     assert response["result"]["protocolVersion"] == "2024-11-05"
     assert response["result"]["serverInfo"]["name"] == "deepseek-mcp"
-    assert response["result"]["serverInfo"]["version"] == "0.5.0"
+    assert response["result"]["serverInfo"]["version"] == "0.5.1"
     assert response["result"]["serverInfo"]["apiKey"] in ("set", "missing")