Output schemas not documented. No visibility into response structure, field types, or pagination details. Tools return API responses but LLMs have no guidance on what fields to expect or how to chain results to downstream tools.

Parameter constraints missing. Parameters like 'similarity_top_k' (range 1-100), 'offset' (pagination), and 'flatFilters' (object structure) lack detailed descriptions of valid ranges, formats, or expected structure. LLMs cannot infer what values are valid.

Tool names contain '_juspay' suffix, which is redundant given the server context. Names like 'list_alerts_juspay' could be 'list_alerts', the domain is already clear from the server. This adds 7+ chars to every name without semantic value.

No error handling or recovery guidance. Tool descriptions do not indicate what errors might occur (auth failure, invalid merchantId, API throttling) or what the LLM should do next. Based on juspay_dashboard_mcp/api/rag_tool.py, auth failures raise generic 'Exception' with minimal context.

Pagination not documented. list_orders_v4_juspay accepts 'offset' and returns results but no indication of total count, next_cursor, or whether results are capped. list_alerts_juspay, list_offers_juspay, list_gateway_scheme_juspay lack pagination details entirely.

Parameter descriptions are minimal. Example: 'task_uid' in alert_details_juspay is described as 'Unique identifier for the alert task' but lacks guidance on format (UUID? numeric? string length). 'merchantId' (admin only) is vague, when is it required vs optional? What permission is needed?