mcpHydroSSH has 10 tools with complete input schemas and descriptions. Naming follows verb_noun convention (ssh_list, ssh_connect, ssh_exec). However, descriptions are terse (avg ~40 chars, well below the 194-char baseline), lacking context on WHEN to use each tool and dependencies between them. Parameter descriptions exist but are minimal. Output schemas are not documented, LLMs cannot predict what fields to expect from responses. Error handling is present but not visible in the source excerpt. No tool annotations (readOnlyHint, destructiveHint) despite clear risk classifications. ssh_add_server and ssh_update_server expose password as a parameter, violating secret-injection pattern.

• Expand tool descriptions to 100 - 200 characters. Include WHEN to use, prerequisites, and what the tool returns. E.g., 'List all configured SSH servers. Call this first to discover available servers before connecting. Returns server ID, name, host, and port.'
• Document output schemas for all tools. Specify what fields each tool returns (e.g., ssh_list_servers returns [{id, name, host, port, username}, ...]). This enables LLMs to chain tools correctly.
• Remove 'password' parameter from ssh_add_server and ssh_update_server. Instead, accept a 'password_env_var' parameter that references an environment variable, or use a separate secure credential store. Document that passwords must be injected server-side.
• Add tool annotations to all tool definitions. Mark ssh_list_servers, ssh_view_config, ssh_get_status, ssh_help with readOnlyHint=true. Mark ssh_connect, ssh_exec, ssh_add_server, ssh_update_server with destructiveHint=true (or idempotentHint=true if safe to retry). Mark ssh_remove_server with destructiveHint=true.
• Add parameter constraints and dependencies to descriptions. E.g., 'privateKeyPath (required if authMethod is "key"; path must exist and be readable)' and 'timeout (optional, milliseconds, 1000 - 300000, default 30000)'.