Loading ToolBench
time-mcp is a simple, single-tool server with clear naming and functional implementation. The tool name 'get_current_time' follows verb-noun convention correctly. The description is concise (101 chars, within the 10-1024 baseline). The input schema is present and typed. However, the tool lacks several production-grade elements: no output schema documentation, no error recovery guidance in error messages, missing parameter constraints (timezone should ideally validate against IANA timezone list or suggest common examples), and no documentation of what happens when an invalid timezone is provided until the code is read. The server lacks composition patterns (only 1 tool, so N/A) and has minimal error classification, errors return plain text rather than structured guidance for recovery. For a utility tool, this is acceptable but not excellent.
Returns the current date and time in UTC or a specific timezone
Missing output schema documentation. The tool returns text content, but the structure (what fields? what format?) is not formally documented. LLMs cannot predict the response shape for downstream composition.
Parameter lacks format constraints and validation hints. The 'timezone' parameter accepts free-form strings with no enum or pattern validation declared in the schema. The description includes examples but no formal constraint. This invites invalid timezone strings and relies on runtime try-catch rather than schema-level validation.
Error messages are plain text without recovery guidance. When an invalid timezone is provided, the error says 'Please use IANA timezone names' but does not guide the LLM to call a discovery tool or list valid options. Error responses should categorize as retryable/user-fixable and offer next steps.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 54 | 2026-07-28+ | v2 |
| 2026-03-09 | D | 50 | - | v1 |
No idempotency hint. The tool is read-only and safe to retry, but this is not declared via tool annotations (readOnlyHint or similar). Without explicit hints, agents may be cautious about retrying on transient failures.