Single tool 'run_command' has a basic schema with two parameters (command, cwd) but critical gaps undermine quality. Description is minimal (11 words, ~60 chars), below the 10 - 1024 char guideline and lacks context on when/why to use it. No parameter descriptions visible in the schema, LLMs cannot infer what 'cwd' means or how it interacts with 'command'. No output schema documented. No error handling guidance. The tool executes arbitrary shell commands (WRITE risk) with no mention of safety constraints, validation, or recovery paths. Schema is present but incomplete; descriptions are missing or trivial.

• Expand tool description to 100 - 200 chars. Example: 'Execute a shell command in a specified working directory. Returns exit code, stdout, and stderr. Use for system tasks, file operations, or script execution. Requires explicit approval for destructive commands.'
• Add description to 'command' parameter: 'The shell command to execute (string). Must be a valid bash/sh command. Avoid shell metacharacters unless intentional; use parameter escaping for user input.'
• Add description to 'cwd' parameter: 'Working directory for command execution (optional, string). If omitted, defaults to the current process directory. Must be an absolute or relative path.'
• Document output schema: 'Returns {exit_code: number, stdout: string, stderr: string, duration_ms: number}. Exit code 0 indicates success; non-zero indicates failure. Stdout and stderr are captured separately for error diagnosis.'
• Add error handling guidance: 'On non-zero exit, return the stderr output and suggest recovery (e.g., 'Command failed with exit code 1. Check syntax or permissions.'). For timeout, suggest retrying with a longer deadline or breaking the command into smaller steps.'