patchllm 0.2.1__py3-none-any.whl → 1.0.0__py3-none-any.whl

patchllm/tui/interface.py

@@ -0,0 +1,703 @@
+from rich.console import Console
+from rich.panel import Panel
+from rich.text import Text
+from rich.markup import escape
+from rich.json import JSON
+from pathlib import Path
+import argparse
+import json
+import os
+import re
+import pprint
+
+from prompt_toolkit import PromptSession
+from prompt_toolkit.history import FileHistory
+from prompt_toolkit.completion import FuzzyCompleter
+from litellm import model_list
+
+from .completer import PatchLLMCompleter
+from ..agent.session import AgentSession
+from ..agent import actions
+from ..interactive.selector import select_files_interactively
+from ..patcher import apply_external_patch
+from ..cli.handlers import handle_scope_management
+from ..scopes.builder import helpers
+from ..utils import write_scopes_to_file
+
+SESSION_FILE_PATH = Path(".patchllm_session.json")
+
+def _print_help():
+    help_text = Text()
+    help_text.append("PatchLLM Agent Commands\n\n", style="bold")
+    help_text.append("Agent Workflow:\n", style="bold cyan")
+    help_text.append("  /task <goal>", style="bold"); help_text.append("\n    ↳ Sets the high-level goal.\n")
+    help_text.append("  /plan", style="bold"); help_text.append("\n    ↳ Generates a plan or opens interactive management if a plan exists.\n")
+    help_text.append("  /ask <question>", style="bold"); help_text.append("\n    ↳ Ask a question about the plan or the code context.\n")
+    help_text.append("  /refine <feedback>", style="bold"); help_text.append("\n    ↳ Refine the plan with new feedback/ideas.\n")
+    help_text.append("  /plan --edit <N> <text>", style="bold"); help_text.append("\n    ↳ Edits step N of the plan.\n")
+    help_text.append("  /plan --rm <N>", style="bold"); help_text.append("\n    ↳ Removes step N from the plan.\n")
+    help_text.append("  /plan --add <text>", style="bold"); help_text.append("\n    ↳ Adds a new step to the end of the plan.\n")
+    help_text.append("  /run", style="bold"); help_text.append("\n    ↳ Executes only the next step in the plan.\n")
+    help_text.append("  /run all", style="bold"); help_text.append("\n    ↳ Executes all remaining steps as a single task.\n")
+    help_text.append("  /skip", style="bold"); help_text.append("\n    ↳ Skips the current step.\n")
+    help_text.append("  /diff [all|filename]", style="bold"); help_text.append("\n    ↳ Shows the full diff for a file or all files.\n")
+    help_text.append("  /approve", style="bold"); help_text.append("\n    ↳ Interactively select and apply changes from the last run.\n")
+    help_text.append("  /retry <feedback>", style="bold"); help_text.append("\n    ↳ Retries the last step with new feedback.\n")
+    help_text.append("  /revert", style="bold"); help_text.append("\n    ↳ Reverts the changes from the last /approve.\n\n")
+    help_text.append("Context Management:\n", style="bold cyan")
+    help_text.append("  /context <scope>\n", style="bold"); help_text.append("    ↳ Sets the context using a scope (e.g., @git:staged).\n")
+    help_text.append("  /scopes\n", style="bold"); help_text.append("        ↳ Opens an interactive menu to create and manage scopes.\n\n")
+    help_text.append("Menu & Session:\n", style="bold cyan")
+    help_text.append("  /show [goal|plan|context|history|step]\n", style="bold"); help_text.append(" ↳ Shows session state.\n")
+    help_text.append("  /settings\n", style="bold"); help_text.append("      ↳ Configure the model and API keys.\n")
+    help_text.append("  /help\n", style="bold"); help_text.append("          ↳ Shows this help message.\n")
+    help_text.append("  /exit\n", style="bold"); help_text.append("          ↳ Exits the agent session.\n")
+    return Panel(help_text, title="Help", border_style="green")
+
+def _display_execution_summary(result, console):
+    if not result:
+        console.print("❌ Step failed to produce a result.", style="red")
+        return
+
+    change_summary = result.get("change_summary")
+    if change_summary:
+        console.print(Panel(Text(change_summary, justify="left"), title="Change Summary", border_style="green", expand=False))
+
+    summary = result.get("summary", {})
+    modified = summary.get("modified", [])
+    created = summary.get("created", [])
+
+    if not modified and not created:
+        if not change_summary:
+            console.print("✅ Step finished, but no file changes were detected.", style="yellow")
+        return
+
+    summary_text = Text()
+    if modified:
+        summary_text.append("Modified:\n", style="bold yellow")
+        for f in modified: summary_text.append(f"  - {f}\n")
+    if created:
+        if modified:
+            summary_text.append("\n")
+        summary_text.append("Created:\n", style="bold green")
+        for f in created: summary_text.append(f"  - {f}\n")
+
+    console.print(Panel(summary_text, title="Proposed File Changes", border_style="cyan", expand=False))
+
+def _save_session(session: AgentSession):
+    with open(SESSION_FILE_PATH, 'w') as f: json.dump(session.to_dict(), f, indent=2)
+
+def _clear_session():
+    if SESSION_FILE_PATH.exists(): os.remove(SESSION_FILE_PATH)
+
+def _run_settings_tui(session: AgentSession, console: Console):
+    """A sub-TUI for managing agent settings."""
+    try:
+        from InquirerPy import prompt
+        from InquirerPy.exceptions import InvalidArgument
+        from InquirerPy.validator import EmptyInputValidator
+    except ImportError:
+        console.print("❌ 'InquirerPy' is required. `pip install 'patchllm[interactive]'`", style="red"); return
+
+    console.print("\n--- Agent Settings ---", style="bold yellow")
+    while True:
+        try:
+            current_model = session.args.model
+            api_keys_count = len(session.api_keys)
+            action_q = {
+                "type": "list", 
+                "name": "action", 
+                "message": "Select a setting to configure:", 
+                "choices": [
+                    f"Change Model (current: {current_model})", 
+                    f"Manage API Keys ({api_keys_count} saved)",
+                    "Back to agent"
+                ], 
+                "border": True, "cycle": False
+            }
+            result = prompt([action_q])
+            action = result.get("action") if result else "Back to agent"
+            
+            if action == "Back to agent": break
+
+            if action.startswith("Change Model"):
+                model_q = {
+                    "type": "fuzzy", 
+                    "name": "model", 
+                    "message": "Fuzzy search for a model:",
+                    "choices": model_list,
+                    "default": current_model
+                }
+                model_r = prompt([model_q])
+                new_model = model_r.get("model") if model_r else None
+                if new_model:
+                    session.args.model = new_model
+                    session.save_settings()
+                    console.print(f"✅ Default model set to '[bold]{new_model}[/bold]'. This will be saved.", style="green")
+
+            elif action.startswith("Manage API Keys"):
+                while True:
+                    saved_keys = list(session.api_keys.keys())
+                    key_choices = ["Add/Update a saved API Key"]
+                    if saved_keys:
+                        key_choices.append("Remove a saved API Key")
+                    key_choices.append("Back")
+                    
+                    key_action_q = {"type": "list", "name": "key_action", "message": "Manage your saved API keys", "choices": key_choices}
+                    key_action_r = prompt([key_action_q])
+                    key_action = key_action_r.get("key_action") if key_action_r else "Back"
+
+                    if key_action == "Back": break
+
+                    if key_action.startswith("Add/Update"):
+                        env_var_q = {"type": "input", "name": "env_var", "message": "Enter the environment variable name (e.g., OPENAI_API_KEY):", "validate": EmptyInputValidator()}
+                        env_var_r = prompt([env_var_q])
+                        env_var_name = env_var_r.get("env_var") if env_var_r else None
+                        if not env_var_name: continue
+
+                        key_q = {"type": "password", "name": "key", "message": f"Enter the value for {env_var_name}:"}
+                        key_r = prompt([key_q])
+                        api_key = key_r.get("key") if key_r else None
+                        if api_key:
+                            session.set_api_key(env_var_name, api_key)
+                            console.print(f"✅ Key '{env_var_name}' has been saved and applied to the current session.", style="green")
+
+                    elif key_action.startswith("Remove"):
+                        remove_q = {"type": "checkbox", "name": "keys", "message": "Select keys to remove:", "choices": saved_keys}
+                        remove_r = prompt([remove_q])
+                        keys_to_remove = remove_r.get("keys") if remove_r else []
+                        if keys_to_remove:
+                            for key in keys_to_remove:
+                                session.remove_api_key(key)
+                            console.print(f"✅ Removed {len(keys_to_remove)} key(s).", style="green")
+
+        except (KeyboardInterrupt, InvalidArgument, IndexError, KeyError, TypeError): break
+    console.print("\n--- Returning to Agent ---", style="bold yellow")
+
+def _edit_string_list_interactive(current_list: list[str], item_name: str, console: Console) -> list[str] | None:
+    """Helper TUI to add/remove items from a simple list of strings."""
+    try:
+        from InquirerPy import prompt
+        from InquirerPy.validator import EmptyInputValidator
+    except ImportError: return None
+
+    edited_list = current_list[:]
+    while True:
+        console.print(Panel(f"[bold]Current {item_name}s:[/]\n" + "\n".join(f"- {i}" for i in edited_list) if edited_list else "  (empty)", expand=False))
+        action_q = {"type": "list", "name": "action", "message": f"Manage {item_name}s", "choices": [f"Add a {item_name}", f"Remove a {item_name}", "Done"], "border": True}
+        action_r = prompt([action_q])
+        action = action_r.get("action") if action_r else "Done"
+
+        if action == "Done": return edited_list
+        if action == f"Add a {item_name}":
+            item_q = {"type": "input", "name": "item", "message": "Enter new item:", "validate": EmptyInputValidator()}
+            item_r = prompt([item_q])
+            new_item = item_r.get("item") if item_r else None
+            if new_item: edited_list.append(new_item)
+        elif action == f"Remove a {item_name}":
+            if not edited_list: console.print(f"No {item_name}s to remove.", style="yellow"); continue
+            remove_q = {"type": "checkbox", "name": "items", "message": "Select items to remove", "choices": edited_list}
+            remove_r = prompt([remove_q])
+            to_remove = remove_r.get("items", []) if remove_r else []
+            edited_list = [item for item in edited_list if item not in to_remove]
+
+def _edit_patterns_interactive(current_list: list[str], pattern_type: str, console: Console) -> list[str] | None:
+    """Helper TUI to add/remove glob patterns, with an option for interactive selection."""
+    try:
+        from InquirerPy import prompt
+        from InquirerPy.validator import EmptyInputValidator
+    except ImportError: return None
+
+    edited_list = current_list[:]
+    while True:
+        console.print(Panel(f"[bold]Current {pattern_type} Patterns:[/]\n" + "\n".join(f"- {i}" for i in edited_list) if edited_list else "  (empty)", expand=False))
+        choices = ["Add pattern manually", "Remove a pattern", "Add from interactive selector", "Done"]
+        action_q = {"type": "list", "name": "action", "message": "Manage patterns", "choices": choices, "border": True}
+        action_r = prompt([action_q])
+        action = action_r.get("action") if action_r else "Done"
+
+        if action == "Done": return edited_list
+        elif action == "Add pattern manually":
+            item_q = {"type": "input", "name": "item", "message": "Enter glob pattern (e.g., 'src/**/*.py'):", "validate": EmptyInputValidator()}
+            item_r = prompt([item_q])
+            new_item = item_r.get("item") if item_r else None
+            if new_item: edited_list.append(new_item)
+        elif action == "Remove a pattern":
+            if not edited_list: console.print("No patterns to remove.", style="yellow"); continue
+            remove_q = {"type": "checkbox", "name": "items", "message": "Select patterns to remove", "choices": edited_list}
+            remove_r = prompt([remove_q])
+            to_remove = remove_r.get("items", []) if remove_r else []
+            edited_list = [item for item in edited_list if item not in to_remove]
+        elif action == "Add from interactive selector":
+            base_path = Path(".").resolve()
+            selected_files = select_files_interactively(base_path)
+            if selected_files:
+                new_patterns = [p.relative_to(base_path).as_posix() for p in selected_files]
+                edited_list.extend(new_patterns)
+                console.print(f"✅ Added {len(new_patterns)} file/folder pattern(s).", style="green")
+
+def _interactive_scope_editor(console: Console, existing_scope: dict | None = None) -> dict | None:
+    """A TUI for creating or editing a single scope dictionary."""
+    try:
+        from InquirerPy import prompt
+    except ImportError:
+        console.print("❌ 'InquirerPy' is required. `pip install 'patchllm[interactive]'`", style="red"); return None
+
+    if existing_scope:
+        scope_data = existing_scope.copy()
+    else: # Default for a new scope
+        scope_data = {"path": ".", "include_patterns": ["**/*"], "exclude_patterns": [], "search_words": [], "urls": [], "exclude_extensions": []}
+
+    while True:
+        console.print(Panel(JSON(json.dumps(scope_data)), title="Current Scope Configuration", border_style="blue"))
+        
+        choices = [
+            f"Edit base path",
+            f"Manage include patterns ({len(scope_data.get('include_patterns', []))})",
+            f"Manage exclude patterns ({len(scope_data.get('exclude_patterns', []))})",
+            f"Manage search keywords ({len(scope_data.get('search_words', []))})",
+            f"Manage URLs ({len(scope_data.get('urls', []))})",
+            f"Manage excluded extensions ({len(scope_data.get('exclude_extensions', []))})",
+            "Save and Return",
+            "Cancel and Discard"
+        ]
+        action_q = {"type": "list", "name": "action", "message": "Select field to edit", "choices": choices, "border": True}
+        action_r = prompt([action_q])
+        action = action_r.get("action") if action_r else "Cancel and Discard"
+
+        if action == "Save and Return": return scope_data
+        if action == "Cancel and Discard": return None
+        
+        if action.startswith("Edit base path"):
+            path_q = {"type": "input", "name": "path", "message": "Enter new base path:", "default": scope_data.get('path', '.')}
+            path_r = prompt([path_q])
+            if path_r and path_r.get("path") is not None: scope_data['path'] = path_r.get("path")
+        elif action.startswith("Manage include patterns"):
+            new_list = _edit_patterns_interactive(scope_data.get('include_patterns', []), "Include", console)
+            if new_list is not None: scope_data['include_patterns'] = new_list
+        elif action.startswith("Manage exclude patterns"):
+            new_list = _edit_patterns_interactive(scope_data.get('exclude_patterns', []), "Exclude", console)
+            if new_list is not None: scope_data['exclude_patterns'] = new_list
+        elif action.startswith("Manage search keywords"):
+            new_list = _edit_string_list_interactive(scope_data.get('search_words', []), "keyword", console)
+            if new_list is not None: scope_data['search_words'] = new_list
+        elif action.startswith("Manage URLs"):
+            new_list = _edit_string_list_interactive(scope_data.get('urls', []), "URL", console)
+            if new_list is not None: scope_data['urls'] = new_list
+        elif action.startswith("Manage excluded extensions"):
+            new_list = _edit_string_list_interactive(scope_data.get('exclude_extensions', []), "extension", console)
+            if new_list is not None: scope_data['exclude_extensions'] = new_list
+
+def _run_scope_management_tui(scopes, scopes_file_path, console):
+    """A sub-TUI for managing scopes, reusing the core handler logic."""
+    try:
+        from InquirerPy import prompt
+        from InquirerPy.validator import EmptyInputValidator
+        from InquirerPy.exceptions import InvalidArgument
+    except ImportError:
+        console.print("❌ 'InquirerPy' is required. `pip install 'patchllm[interactive]'`", style="red"); return
+
+    console.print("\n--- Scope Management ---", style="bold yellow")
+    while True:
+        try:
+            choices = ["List scopes", "Show a scope", "Add a new scope", "Update a scope", "Remove a scope", "Back to agent"]
+            action_q = {"type": "list", "name": "action", "message": "Select an action:", "choices": choices, "border": True, "cycle": False}
+            result = prompt([action_q])
+            action = result.get("action") if result else "Back to agent"
+            if action == "Back to agent": break
+
+            # --- MODIFICATION: Create a base Namespace with all expected keys ---
+            base_args = argparse.Namespace(
+                list_scopes=False, show_scope=None, add_scope=None,
+                remove_scope=None, update_scope=None
+            )
+
+            if action == "List scopes":
+                base_args.list_scopes = True
+                handle_scope_management(base_args, scopes, scopes_file_path, None)
+            
+            elif action == "Show a scope":
+                if not scopes: console.print("No scopes to show.", style="yellow"); continue
+                scope_q = {"type": "fuzzy", "name": "scope", "message": "Which scope to show?", "choices": sorted(scopes.keys())}
+                scope_r = prompt([scope_q])
+                if scope_r and scope_r.get("scope"):
+                    base_args.show_scope = scope_r.get("scope")
+                    handle_scope_management(base_args, scopes, scopes_file_path, None)
+            
+            elif action == "Remove a scope":
+                if not scopes: console.print("No scopes to remove.", style="yellow"); continue
+                scope_q = {"type": "fuzzy", "name": "scope", "message": "Which scope to remove?", "choices": sorted(scopes.keys())}
+                scope_r = prompt([scope_q])
+                if scope_r and scope_r.get("scope"):
+                    base_args.remove_scope = scope_r.get("scope")
+                    handle_scope_management(base_args, scopes, scopes_file_path, None)
+            
+            elif action == "Add a new scope":
+                name_q = {"type": "input", "name": "name", "message": "Enter name for the new scope:", "validate": EmptyInputValidator()}
+                name_r = prompt([name_q])
+                scope_name = name_r.get("name") if name_r else None
+                if not scope_name: continue
+                if scope_name in scopes: console.print(f"❌ Scope '{scope_name}' already exists.", style="red"); continue
+                
+                new_scope_data = _interactive_scope_editor(console, existing_scope=None)
+                if new_scope_data:
+                    scopes[scope_name] = new_scope_data
+                    write_scopes_to_file(scopes_file_path, scopes)
+                    console.print(f"✅ Scope '{scope_name}' created.", style="green")
+
+            elif action == "Update a scope":
+                if not scopes: console.print("No scopes to update.", style="yellow"); continue
+                scope_q = {"type": "fuzzy", "name": "scope", "message": "Which scope to update?", "choices": sorted(scopes.keys())}
+                scope_r = prompt([scope_q])
+                scope_name = scope_r.get("scope") if scope_r else None
+                if not scope_name: continue
+                
+                updated_scope_data = _interactive_scope_editor(console, existing_scope=scopes[scope_name])
+                if updated_scope_data:
+                    scopes[scope_name] = updated_scope_data
+                    write_scopes_to_file(scopes_file_path, scopes)
+                    console.print(f"✅ Scope '{scope_name}' updated.", style="green")
+
+        except (KeyboardInterrupt, InvalidArgument, IndexError, KeyError, TypeError): break
+    console.print("\n--- Returning to Agent ---", style="bold yellow")
+
+def _run_plan_management_tui(session: AgentSession, console: Console):
+    """A sub-TUI for interactively managing the execution plan."""
+    try:
+        from InquirerPy import prompt
+        from InquirerPy.validator import EmptyInputValidator
+        from InquirerPy.exceptions import InvalidArgument
+    except ImportError:
+        console.print("❌ 'InquirerPy' is required. `pip install 'patchllm[interactive]'`", style="red"); return
+
+    console.print("\n--- Interactive Plan Management ---", style="bold yellow")
+    while True:
+        try:
+            if not session.plan:
+                console.print("The plan is now empty.", style="yellow")
+                break
+
+            choices = [f"{i+1}. {step}" for i, step in enumerate(session.plan)]
+            
+            action_q = {
+                "type": "list", "name": "action", "message": "Select a step to manage or an action:",
+                "choices": choices + ["Add a new step", "Reorder steps", "Done"],
+                "border": True, "cycle": False,
+                "long_instruction": "Use arrow keys. Select a step to Edit/Remove it."
+            }
+            result = prompt([action_q])
+            action_choice = result.get("action") if result else "Done"
+
+            if action_choice == "Done": break
+            
+            if action_choice == "Add a new step":
+                add_q = {"type": "input", "name": "text", "message": "Enter the new step instruction:", "validate": EmptyInputValidator()}
+                add_r = prompt([add_q])
+                if add_r and add_r.get("text"):
+                    session.add_plan_step(add_r.get("text"))
+                    console.print("✅ Step added to the end of the plan.", style="green")
+
+            elif action_choice == "Reorder steps":
+                if len(session.plan) < 2:
+                    console.print("⚠️ Not enough steps to reorder.", style="yellow"); continue
+                
+                reorder_choices = [f"{i+1}. {step}" for i, step in enumerate(session.plan)]
+                
+                from_q = {
+                    "type": "list", "name": "from", "message": "Select the step to move:",
+                    "choices": reorder_choices, "cycle": False
+                }
+                from_r = prompt([from_q])
+                if not from_r or not from_r.get("from"): continue
+                from_index = int(from_r.get("from").split('.')[0]) - 1
+
+                to_choices = [f"Move to position {i+1}" for i in range(len(session.plan))]
+                to_q = {
+                    "type": "list", "name": "to", "message": f"Where should '{session.plan[from_index]}' move?",
+                    "choices": to_choices, "cycle": False
+                }
+                to_r = prompt([to_q])
+                if not to_r or not to_r.get("to"): continue
+                to_index = int(re.search(r'\d+', to_r.get("to")).group()) - 1
+                
+                item_to_move = session.plan.pop(from_index)
+                session.plan.insert(to_index, item_to_move)
+                console.print(f"✅ Step moved from position {from_index + 1} to {to_index + 1}.", style="green")
+
+            else:
+                step_index = int(action_choice.split('.')[0]) - 1
+                step_text = session.plan[step_index]
+
+                edit_or_rm_q = {
+                    "type": "list", "name": "sub_action", "message": f"Step {step_index + 1}: {step_text}",
+                    "choices": ["Edit", "Remove", "Cancel"]
+                }
+                edit_or_rm_r = prompt([edit_or_rm_q])
+                sub_action = edit_or_rm_r.get("sub_action") if edit_or_rm_r else "Cancel"
+
+                if sub_action == "Edit":
+                    edit_q = {"type": "input", "name": "text", "message": "Enter the new instruction:", "default": step_text, "validate": EmptyInputValidator()}
+                    edit_r = prompt([edit_q])
+                    if edit_r and edit_r.get("text"):
+                        session.edit_plan_step(step_index + 1, edit_r.get("text"))
+                        console.print(f"✅ Step {step_index + 1} updated.", style="green")
+                
+                elif sub_action == "Remove":
+                    session.remove_plan_step(step_index + 1)
+                    console.print(f"✅ Step {step_index + 1} removed.", style="green")
+        
+        except (KeyboardInterrupt, InvalidArgument, IndexError, KeyError, TypeError): break
+
+    _save_session(session)
+    console.print("\n--- Returning to Agent ---", style="bold yellow")
+
+def run_tui(args, scopes, recipes, scopes_file_path):
+    console = Console()
+    session = AgentSession(args, scopes, recipes)
+
+    if SESSION_FILE_PATH.exists():
+        if console.input("Found saved session. [bold]Resume?[/bold] (Y/n) ").lower() in ['y', 'yes', '']:
+            try:
+                with open(SESSION_FILE_PATH, 'r') as f: session.from_dict(json.load(f))
+                console.print("✅ Session resumed.", style="green")
+            except Exception as e: console.print(f"⚠️ Could not resume session: {e}", style="yellow"); _clear_session()
+        else: _clear_session()
+
+    completer = PatchLLMCompleter(scopes=session.scopes)
+    prompt_session = PromptSession(history=FileHistory(Path("~/.patchllm_history").expanduser()))
+
+    console.print("🤖 Welcome to the PatchLLM Agent. Type `/` and [TAB] for commands. `/exit` to quit.", style="bold blue")
+
+    try:
+        while True:
+            completer.set_session_state(
+                has_goal=bool(session.goal),
+                has_plan=bool(session.plan),
+                has_pending_changes=bool(session.last_execution_result),
+                can_revert=bool(session.last_revert_state),
+                has_context=bool(session.context)
+            )
+            
+            text = prompt_session.prompt(">>> ", completer=FuzzyCompleter(completer)).strip()
+            if not text: continue
+            
+            command, _, arg_string = text.partition(' ')
+            command = command.lower()
+            
+            if command == '/exit': _clear_session(); break
+            elif command == '/help': console.print(_print_help())
+            elif command == '/task':
+                session.set_goal(arg_string); console.print("✅ Goal set.", style="green"); _save_session(session)
+            
+            elif command == '/ask':
+                if not session.plan and not session.context:
+                    console.print("❌ No plan or context to ask about. Use `/context` to load files or `/plan` to generate a plan.", style="red"); continue
+                if not arg_string: console.print("❌ Please provide a question.", style="red"); continue
+                with console.status("[cyan]Asking assistant..."): response = session.ask_question(arg_string)
+                if response:
+                    console.print(Panel(response, title="Assistant's Answer", border_style="blue"))
+                else: console.print("❌ Failed to get a response.", style="red")
+
+            elif command == '/refine':
+                if not session.plan: console.print("❌ No plan to refine. Generate one with `/plan` first.", style="red"); continue
+                if not arg_string: console.print("❌ Please provide feedback or an idea.", style="red"); continue
+                with console.status("[cyan]Refining plan..."): success = session.refine_plan(arg_string)
+                if success:
+                    console.print(Panel("\n".join(f"{i+1}. {s}" for i, s in enumerate(session.plan)), title="Refined Execution Plan", border_style="magenta"))
+                    _save_session(session)
+                else: console.print("❌ Failed to refine the plan.", style="red")
+
+            elif command == '/plan':
+                if not arg_string:
+                    if not session.goal and not session.plan:
+                        console.print("❌ No goal set. Set one with `/task <your goal>`.", style="red"); continue
+                    
+                    if session.plan:
+                        _run_plan_management_tui(session, console)
+                        if session.plan:
+                            console.print(Panel("\n".join(f"{i+1}. {s}" for i, s in enumerate(session.plan)), title="Current Execution Plan", border_style="magenta"))
+                        continue
+
+                    with console.status("[cyan]Generating plan..."): success = session.create_plan()
+                    if success:
+                        console.print(Panel("\n".join(f"{i+1}. {s}" for i, s in enumerate(session.plan)), title="Execution Plan", border_style="magenta")); _save_session(session)
+                    else: console.print("❌ Failed to generate a plan.", style="red")
+                else:
+                    if not session.plan:
+                        console.print("❌ No plan to manage. Generate one with `/plan` first.", style="red"); continue
+                    
+                    edit_match = re.match(r"--edit\s+(\d+)\s+(.*)", arg_string, re.DOTALL)
+                    rm_match = re.match(r"--rm\s+(\d+)", arg_string)
+                    add_match = re.match(r"--add\s+(.*)", arg_string, re.DOTALL)
+
+                    if edit_match:
+                        step_num, new_text = int(edit_match.group(1)), edit_match.group(2)
+                        if session.edit_plan_step(step_num, new_text):
+                            console.print(f"✅ Step {step_num} updated.", style="green"); _save_session(session)
+                        else: console.print(f"❌ Invalid step number: {step_num}.", style="red")
+                    elif rm_match:
+                        step_num = int(rm_match.group(1))
+                        if session.remove_plan_step(step_num):
+                            console.print(f"✅ Step {step_num} removed.", style="green"); _save_session(session)
+                        else: console.print(f"❌ Invalid step number: {step_num}.", style="red")
+                    elif add_match:
+                        new_text = add_match.group(1)
+                        session.add_plan_step(new_text)
+                        console.print("✅ New step added to the end of the plan.", style="green"); _save_session(session)
+                    else:
+                        console.print(f"❌ Unknown argument for /plan: '{arg_string}'. Use --edit, --rm, or --add.", style="red")
+
+                    console.print(Panel("\n".join(f"{i+1}. {s}" for i, s in enumerate(session.plan)), title="Updated Execution Plan", border_style="magenta"))
+            
+            elif command == '/run':
+                result = None
+                if session.plan:
+                    if session.current_step >= len(session.plan):
+                        console.print("✅ Plan complete.", style="green")
+                        continue
+                    
+                    if arg_string == 'all':
+                        remaining_count = len(session.plan) - session.current_step
+                        console.print(f"\n--- Executing all {remaining_count} remaining steps ---", style="bold yellow")
+                        with console.status("[cyan]Agent is working..."):
+                            result = session.run_all_remaining_steps()
+                    else:
+                        console.print(f"\n--- Executing Step {session.current_step + 1}/{len(session.plan)} ---", style="bold yellow")
+                        with console.status("[cyan]Agent is working..."):
+                            result = session.run_next_step()
+                else:
+                    if not session.goal:
+                        console.print("❌ No plan or goal is set. Use `/task <goal>` to set a goal first.", style="red")
+                        continue
+                    
+                    console.print("\n--- Executing Goal Directly (No Plan) ---", style="bold yellow")
+                    with console.status("[cyan]Agent is working..."):
+                        result = session.run_goal_directly()
+
+                _display_execution_summary(result, console)
+                if result:
+                    console.print("✅ Preview ready. Use `/diff` to review.", style="green")
+
+            elif command == '/skip':
+                if not session.plan: console.print("❌ No plan to skip from.", style="red"); continue
+                if session.skip_step():
+                    console.print(f"✅ Step {session.current_step} skipped. Now at step {session.current_step + 1}.", style="green")
+                    _save_session(session)
+                else:
+                    console.print("✅ Plan already complete. Nothing to skip.", style="green")
+
+            elif command == '/diff':
+                if not session.last_execution_result or not session.last_execution_result.get("diffs"): console.print("❌ No diff to display.", style="red"); continue
+                diffs = session.last_execution_result["diffs"]
+                if arg_string and arg_string != 'all': diffs = [d for d in diffs if Path(d['file_path']).name == arg_string]
+                for diff in diffs: console.print(Panel(diff["diff_text"], title=f"Diff: {Path(diff['file_path']).name}", border_style="yellow"))
+
+            elif command == '/approve':
+                if not session.last_execution_result: console.print("❌ No changes to approve.", style="red"); continue
+                
+                try:
+                    from InquirerPy import prompt
+                    summary = session.last_execution_result.get("summary", {})
+                    all_files = summary.get("modified", []) + summary.get("created", [])
+                    
+                    if not all_files:
+                        console.print("✅ No file changes were proposed to approve.", style="yellow")
+                        session.last_execution_result = None
+                        continue
+
+                    approve_q = {
+                        "type": "checkbox", "name": "files", "message": "Select the changes you wish to apply:",
+                        "choices": all_files, "validate": lambda r: len(r) > 0,
+                        "invalid_message": "You must select at least one file to apply.",
+                        "transformer": lambda r: f"{len(r)} file(s) selected."
+                    }
+                    result = prompt([approve_q])
+                    files_to_approve = result.get("files") if result else []
+
+                    if not files_to_approve:
+                        console.print("Approval cancelled.", style="yellow"); continue
+
+                    with console.status("[cyan]Applying..."):
+                        is_full_approval = session.approve_changes(files_to_approve)
+                    
+                    if is_full_approval:
+                        console.print("✅ All changes applied. Moving to the next step.", style="green")
+                    else:
+                        console.print("✅ Partial changes applied.", style="green")
+                        console.print("👉 Use `/retry <feedback>` to fix the remaining files, or `/skip` to move on.", style="cyan")
+                    
+                    _save_session(session)
+
+                except (KeyboardInterrupt, ImportError):
+                    console.print("Approval cancelled.", style="yellow")
+
+            elif command == '/retry':
+                if not session.last_execution_result:
+                    console.print("❌ Nothing to retry.", style="red")
+                    continue
+                if not arg_string:
+                    console.print("❌ Please provide feedback for the retry.", style="red")
+                    continue
+                
+                console.print("\n--- Retrying ---", style="bold yellow")
+                with console.status("[cyan]Agent is working..."):
+                    result = session.retry_step(arg_string)
+                _display_execution_summary(result, console)
+
+            elif command == '/revert':
+                if not session.last_revert_state: console.print("❌ No approved changes to revert.", style="red"); continue
+                with console.status("[cyan]Reverting changes..."): success = session.revert_last_approval()
+                if success:
+                    console.print("✅ Last approved changes have been reverted.", style="green"); _save_session(session)
+                else: console.print("❌ Failed to revert changes.", style="red")
+
+            elif command == '/show':
+                if arg_string == 'goal':
+                    if not session.goal: console.print("No goal set.", style="yellow")
+                    else: console.print(Panel(escape(session.goal), title="Current Goal", border_style="blue"))
+                elif arg_string == 'plan':
+                    if not session.plan: console.print("No plan exists.", style="yellow")
+                    else: console.print(Panel("\n".join(f"{i+1}. {s}" for i, s in enumerate(session.plan)), title="Execution Plan", border_style="magenta"))
+                elif arg_string == 'step':
+                    if not session.plan:
+                        console.print("No plan exists.", style="yellow")
+                    elif session.current_step >= len(session.plan):
+                        console.print("✅ Plan is complete.", style="green")
+                    else:
+                        step_text = Text()
+                        step_text.append(f"Current Step ({session.current_step + 1}/{len(session.plan)}):\n", style="bold green")
+                        step_text.append(f"  -> {escape(session.plan[session.current_step])}\n")
+                        if session.current_step + 1 < len(session.plan):
+                            step_text.append(f"\nNext Step ({session.current_step + 2}/{len(session.plan)}):\n", style="bold blue")
+                            step_text.append(f"  -> {escape(session.plan[session.current_step + 1])}")
+                        console.print(Panel(step_text, title="Current Step", border_style="magenta"))
+                elif arg_string == 'context':
+                    if not session.context_files: console.print("Context is empty.", style="yellow")
+                    else:
+                        tree = helpers.generate_source_tree(Path(".").resolve(), session.context_files)
+                        console.print(Panel(tree, title="Context Tree", border_style="cyan"))
+                elif arg_string == 'history':
+                    if not session.action_history: console.print("No actions recorded yet.", style="yellow")
+                    else:
+                        history_text = Text()
+                        for i, entry in enumerate(session.action_history): history_text.append(f"{i+1}. {escape(entry)}\n")
+                        console.print(Panel(history_text, title="Session History", border_style="blue"))
+                else:
+                    console.print("Usage: /show [goal|plan|context|history|step]", style="yellow")
+
+            elif command == '/context':
+                with console.status("[cyan]Building..."): summary = session.load_context_from_scope(arg_string)
+                console.print(Panel(summary, title="Context Summary", border_style="cyan")); _save_session(session)
+            
+            elif command == '/scopes':
+                _run_scope_management_tui(session.scopes, scopes_file_path, console)
+                session.reload_scopes(scopes_file_path)
+            
+            elif command == '/settings':
+                _run_settings_tui(session, console)
+            
+            else:
+                console.print(f"Unknown command: '{text}'.", style="yellow")
+    except (KeyboardInterrupt, EOFError): console.print()
+    except Exception as e: console.print(f"An unexpected error occurred: {e}", style="bold red")
+    console.print("\n👋 Exiting agent session. Goodbye!", style="yellow")