feat: implement tweaks plan - modals, smart shell, spinner, /models, debug log, skills

Phase 1: Permission modal dialog, session resume modal, HistoryInput with
up/down arrow cycling, remove "You:" echo from chat log, LLM client cleanup
on unmount.

Phase 2: Smart shell auto-approve using allowed/denied command lists from
ToolsConfig, animated thinking spinner with live token count in status bar.

Phase 3: /models slash command (list + switch), CLI directory positional
argument, JSONL debug logger with rotation.

Phase 4: Skills system with SkillsManager, load_skill LLM tool, /skills
listing, skill invocation via slash commands, system prompt integration.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

app/services/permissions.py

@@ -2,10 +2,12 @@
 
 from __future__ import annotations
 
+import json
 import logging
+import shlex
 from collections.abc import Awaitable, Callable
 
-from app.models.config import PermissionsConfig
+from app.models.config import PermissionsConfig, ToolsConfig
 
 logger = logging.getLogger(__name__)
 
@@ -24,8 +26,13 @@ class PermissionsService:
     shows a modal dialog. Without a callback, unlisted tools are denied.
     """
 
-    def __init__(self, config: PermissionsConfig) -> None:
+    def __init__(
+        self,
+        config: PermissionsConfig,
+        tools_config: ToolsConfig | None = None,
+    ) -> None:
         self.config = config
+        self._tools_config = tools_config
         self._prompt_callback: PromptCallback | None = None
 
     def set_prompt_callback(self, callback: PromptCallback) -> None:
@@ -36,9 +43,19 @@ class PermissionsService:
         """
         self._prompt_callback = callback
 
-    async def check(self, tool_name: str, description: str = "") -> bool:
+    async def check(
+        self,
+        tool_name: str,
+        description: str = "",
+        arguments: str = "",
+    ) -> bool:
         """Check if a tool is permitted to run.
 
+        Args:
+            tool_name: Name of the tool to check.
+            description: Human-readable description for the prompt.
+            arguments: Raw JSON arguments string (used for shell-aware checks).
+
         Returns:
             True if permitted, False if denied.
         """
@@ -50,6 +67,12 @@ class PermissionsService:
             logger.debug("Tool '%s' is auto-approved", tool_name)
             return True
 
+        # Shell-aware: check allowed/denied command lists
+        if tool_name == "run_command" and self._tools_config is not None:
+            result = self._check_shell_command(arguments)
+            if result is not None:
+                return result
+
         # Prompt user via callback (TUI modal, etc.)
         if self._prompt_callback is not None:
             return await self._prompt_callback(tool_name, description)
@@ -57,3 +80,41 @@ class PermissionsService:
         # No callback set — deny by default (safe fallback)
         logger.warning("Tool '%s' requires approval but no prompt callback set — denied", tool_name)
         return False
+
+    def _check_shell_command(self, arguments: str) -> bool | None:
+        """Check shell command against allowed/denied lists.
+
+        Returns True (allow), False (deny), or None (fall through to prompt).
+        """
+        shell_config = self._tools_config.shell  # type: ignore[union-attr]
+
+        try:
+            cmd = json.loads(arguments).get("command", "")
+        except (json.JSONDecodeError, AttributeError):
+            return None  # can't parse, fall through to prompt
+
+        try:
+            base_cmd = shlex.split(cmd)[0]
+        except (ValueError, IndexError):
+            return None
+
+        # Denied commands: prefix match on full command string
+        for denied in shell_config.denied_commands:
+            if cmd.startswith(denied):
+                logger.info("Shell command '%s' matches denied prefix '%s'", cmd, denied)
+                return False
+
+        # Allowed commands: base executable match
+        if shell_config.allowed_commands:
+            if base_cmd in shell_config.allowed_commands:
+                logger.debug(
+                    "Shell command '%s' auto-approved (base '%s' in allowed list)",
+                    cmd,
+                    base_cmd,
+                )
+                return True
+            # Base command NOT in allowed list — fall through to prompt
+            return None
+
+        # No allowed list configured — fall through to prompt
+        return None