Fix readme and cli err on linux

Author: h3xxit

plugins/communication_protocols/cli/README.md

@@ -179,8 +179,6 @@ Some commands work across platforms:
 
 - Commands execute in isolated subprocesses with controlled environment
 - Environment variables provide secure credential passing
-- Working directory restrictions limit file system access  
-- UTCP_ARG placeholders prevent command injection
 - Previous command outputs should be used carefully to avoid injection
 - Commands should use platform-appropriate syntax
 

plugins/communication_protocols/cli/src/utcp_cli/cli_communication_protocol.py

@@ -174,18 +174,14 @@ async def register_manual(self, caller, manual_call_template: CallTemplate) -> R
         )
 
         try:
-            # For discovery, use the first command only
-            first_command = manual_call_template.commands[0]
-            
+            # Execute commands using the same approach as call_tool but with no arguments
             env = self._prepare_environment(manual_call_template)
-            # Parse command string into proper arguments
-            # Use posix=False on Windows, posix=True on Unix-like systems
-            command = shlex.split(first_command.command, posix=(os.name != 'nt'))
-
-            self._log_info(f"Executing command for tool discovery: {' '.join(command)}")
+            shell_script = self._build_combined_shell_script(manual_call_template.commands, {})
+            
+            self._log_info(f"Executing shell script for tool discovery from provider '{manual_call_template.name}'")
 
-            stdout, stderr, return_code = await self._execute_command(
-                command,
+            stdout, stderr, return_code = await self._execute_shell_script(
+                shell_script,
                 env,
                 timeout=30.0,
                 working_dir=manual_call_template.working_dir,
@@ -196,14 +192,14 @@ async def register_manual(self, caller, manual_call_template: CallTemplate) -> R
 
             if not output.strip():
                 self._log_info(
-                    f"No output from command '{manual_call_template.command_name}'"
+                    f"No output from commands for CLI provider '{manual_call_template.name}'"
                 )
                 return RegisterManualResult(
                     success=False,
                     manual_call_template=manual_call_template,
                     manual=UtcpManual(manual_version="0.0.0", tools=[]),
                     errors=[
-                        f"No output from discovery command for CLI provider '{manual_call_template.name}'"
+                        f"No output from discovery commands for CLI provider '{manual_call_template.name}'"
                     ],
                 )
 
@@ -415,8 +411,14 @@ def _parse_combined_output(self, stdout: str, stderr: str, return_code: int, com
         Returns:
             Final output from script (already filtered by append_to_final_output)
         """
-        # Use stdout if successful, stderr if failed
-        output = stdout if return_code == 0 else stderr
+        # Platform-specific output handling
+        if os.name == 'nt':
+            # Windows (PowerShell): Use stdout on success, stderr on failure
+            output = stdout if return_code == 0 else stderr
+        else:
+            # Unix (Bash): Our script captures everything and echoes to stdout
+            # So we always use stdout first, fallback to stderr if stdout is empty
+            output = stdout if stdout.strip() else stderr
 
         if not output.strip():
             self._log_info(f"CLI tool '{tool_name}' produced no output")