Premium Email Management MCP Server for Cloud Mail Flare (Hardened CLI Wrapper). Provides tools to login, manage users, read emails, and configure settings via a sanitized CLI interface.
Two tools with significant definition gaps. cmf_execute is a monolithic command dispatcher with 22 subcommands crammed into one tool, violating single-responsibility principle. Schema is present but overly complex with many optional parameters that create ambiguity. Descriptions are generic and lack actionable guidance. cmf_list_commands is a discovery tool with minimal description. Neither tool follows verb_noun naming conventions (cmf_execute, cmf_list_commands are opaque). Parameter descriptions exist but lack constraint details (e.g., timeout/interval format hints buried in pattern regex, not in description text). No output schemas documented. Error handling guidance absent.
Primary tool to interact with Cloud Mail Flare. Use this to login, manage users, read emails, and configure settings. Subcommands: auth login, auth logout, auth health, user list, user create, user get, user update, user delete, inbox list, inbox get, inbox wait, inbox action, settings get, settings set, apikey list, apikey create, apikey revoke, apikey verify, system dashboard, system cleanup, system benchmark, browser openrouter-signup, browser status
Discover available Cloud Mail Flare subcommands and their parameters.
cmf_execute violates single-responsibility principle: 22 subcommands (auth, user, inbox, settings, apikey, system, browser) bundled into one tool. LLMs cannot infer intent from name alone. Should split into separate tools: execute_auth, execute_user, execute_inbox, etc.
Tool names lack action verbs. 'cmf_execute' and 'cmf_list_commands' are opaque prefixes. Should be 'execute_cloudmail_command' or split into domain-specific verbs like 'authenticate_user', 'list_emails', 'create_user'. Current names force LLMs to read full description to understand purpose.
No output schemas documented. cmf_execute description lists 22 subcommands but does not specify what each returns. LLMs cannot plan downstream calls or extract required fields (e.g., does 'user list' return user_id, email, username?). Violates pattern:tool requirement.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 38 | <=2025-11-25 | v2 |
Parameter descriptions lack constraint details. 'timeout' and 'interval' have regex patterns (^\d+[smhd]?) but descriptions do not explain format. Should state: 'Timeout in seconds, minutes, hours, or days (e.g., 60s, 5m, 2h, 1d)'. LLMs cannot read JSON Schema patterns, they rely on description text.
No error handling guidance. If 'auth login' fails (invalid credentials, network error, rate limit), the tool provides no recovery hint. Should return: 'Authentication failed: invalid email or password. Verify credentials and retry.' Agents need actionable error messages.