Single tool 'search-logs' with reasonable schema but significant gaps in parameter descriptions, no output schema documentation, and lack of error guidance. The tool description is minimal (23 chars) and does not explain WHEN to use it, WHAT it returns, or dependencies. Parameters have descriptions but lack clarity on constraints and format details. Error handling exists but does not guide recovery. Overall, this is a fair-to-poor definition that would require substantial improvement for production use.
search logs from datadog
Tool description is too short (23 chars) and lacks context. Does not explain WHEN to call this tool, WHAT it returns, or any prerequisites. Should be 50-200 chars with clear purpose statement.
Output schema is not documented. The tool returns a JSON response with 'content', 'nextCursor', and 'isError' fields, but the tool description does not specify this structure or document what fields the LLM should expect. This violates the documented output schema requirement.
Parameter 'filterQuery' description lacks format constraints. It mentions 'For more information, see https://...' but does not explain what a query string is or provide usage examples inline. LLMs cannot access external links reliably.
Parameter 'cursor' description is ambiguous about when to use it and how pagination works. It mentions both 'nextCursor' and 'meta.page.after' without clearly stating which is the canonical return field or how limits are enforced. This will confuse agents about pagination.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 54 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 38 | - | v1 |
Error handling does not guide recovery. The catch block returns generic error messages without categorizing them as retryable, user-fixable, or fatal. An LLM receiving an error has no guidance on next steps.
No result limit or pagination guidance documented. The tool does not state whether results are capped, what the default limit is, or how to request more items. Large result sets could exhaust context.
Code comment in handler references 'cline' compatibility as justification for returning full API response instead of just log data. This is an implementation detail that leaks into the code and suggests the output schema was not designed to match agent needs.