add --fast flag

holmes/config.py

@@ -304,7 +304,10 @@ def get_runbook_catalog() -> Optional[RunbookCatalog]:
         return runbook_catalog
 
     def create_console_tool_executor(
-        self, dal: Optional["SupabaseDal"], refresh_status: bool = False
+        self,
+        dal: Optional["SupabaseDal"],
+        refresh_status: bool = False,
+        disable_todos: bool = False,
     ) -> ToolExecutor:
         """
         Creates a ToolExecutor instance configured for CLI usage. This executor manages the available tools
@@ -318,6 +321,13 @@ def create_console_tool_executor(
         cli_toolsets = self.toolset_manager.list_console_toolsets(
             dal=dal, refresh_status=refresh_status
         )
+
+        # Disable the core_investigation toolset (which contains TodoWriteTool) when in fast mode
+        if disable_todos:
+            cli_toolsets = [
+                ts for ts in cli_toolsets if ts.name != "core_investigation"
+            ]
+
         return ToolExecutor(cli_toolsets)
 
     def create_tool_executor(self, dal: Optional["SupabaseDal"]) -> ToolExecutor:
@@ -343,8 +353,11 @@ def create_console_toolcalling_llm(
         dal: Optional["SupabaseDal"] = None,
         refresh_toolsets: bool = False,
         tracer=None,
+        disable_todos: bool = False,
     ) -> "ToolCallingLLM":
-        tool_executor = self.create_console_tool_executor(dal, refresh_toolsets)
+        tool_executor = self.create_console_tool_executor(
+            dal, refresh_toolsets, disable_todos
+        )
         from holmes.core.tool_calling_llm import ToolCallingLLM
 
         return ToolCallingLLM(

holmes/core/prompt.py

@@ -54,13 +54,17 @@ def build_initial_ask_messages(
         runbooks: Optional runbook catalog
         system_prompt_additions: Optional additional system prompt content
     """
+    # Check if todos are enabled by looking for core_investigation toolset
+    has_todos = any(ts.name == "core_investigation" for ts in tool_executor.toolsets)
+
     # Load and render system prompt internally
     system_prompt_template = "builtin://generic_ask.jinja2"
     template_context = {
         "toolsets": tool_executor.toolsets,
         "runbooks": runbooks or {},
         "system_prompt_additions": system_prompt_additions or "",
         "investigation_id": investigation_id,
+        "has_todos": has_todos,
     }
     system_prompt_rendered = load_and_render_prompt(
         system_prompt_template, template_context

holmes/main.py

@@ -215,6 +215,11 @@ def ask(
         "--system-prompt-additions",
         help="Additional content to append to the system prompt",
     ),
+    fast: bool = typer.Option(
+        False,
+        "--fast",
+        help="Fast mode: disables todos tool for quicker responses",
+    ),
 ):
     """
     Ask any question and answer using available tools
@@ -252,6 +257,7 @@ def ask(
         dal=None,  # type: ignore
         refresh_toolsets=refresh_toolsets,  # flag to refresh the toolset status
         tracer=tracer,
+        disable_todos=fast,  # disable todos tool when --fast is used
     )
 
     if prompt_file and prompt:

holmes/plugins/prompts/_general_instructions.jinja2

@@ -51,6 +51,7 @@
 * For any question, try to make the answer specific to the user's cluster.
 ** For example, if asked to port forward, find out the app or pod port (kubectl describe) and provide a port forward command specific to the user's question
 
+{% if has_todos %}
 # MANDATORY Task Management
 
 * You MUST use the TodoWrite tool for ANY investigation requiring multiple steps
@@ -62,6 +63,7 @@
 * Follow ALL tasks in your plan - don't skip any tasks
 * Use task management to ensure you don't miss important investigation steps
 * If you discover additional steps during investigation, add them to your task list using TodoWrite
+{% endif %}
 
 # Tool/function calls