The 'wait' tool has a clear action verb in its name and a reasonable description, but lacks depth in parameter guidance and output schema documentation. The tool definition is explicitly visible in src/index.ts with proper schema registration. However, the descriptions are minimal and do not follow LLM-optimization patterns. The single parameter has type information but lacks actionable formatting guidance or use-case clarity. No output schema is documented, forcing LLMs to infer response structure. Error handling returns structured CallToolResult objects but lacks recovery guidance.
Waits for a specified duration in seconds.
Tool description is vague about use case and constraints. 'Waits for a specified duration in seconds' lacks context for when an LLM should call this tool or how long delays are acceptable. No mention of maximum wait duration, blocking behavior, or typical scenarios (e.g., rate limiting, polling intervals).
Output schema is not documented. The code returns CallToolResult with a text message ('Successfully waited for X seconds'), but LLMs have no formal schema definition for what fields to expect or how to parse the response for downstream tool chaining.
Parameter description is bare. 'The number of seconds to wait' provides minimal guidance. Missing: minimum/maximum bounds (is 0.001s allowed? Is 86400s reasonable?), precision expectations (integer vs float behavior), and whether fractional seconds are supported. Zod schema uses .positive() constraint but this is not surfaced to LLMs via description text.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 10 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 42 | - | v1 |
No error recovery guidance. Error responses return isError:true with a message but do not classify the error (retryable, user-fixable, fatal) or suggest next steps. If a timeout occurs or an invalid duration is passed, the LLM has no directive.
Naming is generic and lacks verb clarity. 'wait' is a valid action verb, but is ambiguous without context. Tools in production systems often use more specific names like 'sleep_seconds', 'delay_execution', or 'rate_limit_wait' to distinguish from other blocking operations. At minimum, the description should explain when to use this over other delay mechanisms.