MCP server for the Quismon monitoring and alerting platform. Provides tools to create and manage monitoring checks, trigger immediate check execution, acknowledge incidents, and create alert rules.
The server provides four tools with visible schemas and descriptions, but lacks production-grade refinement. All tools have input schemas defined in the handler code and basic descriptions. However, parameter descriptions are minimal or missing context, output schemas are not documented, and error handling lacks guidance. Tool naming follows verb-noun convention (create_*, trigger_*, acknowledge_*) which is good. The schema definitions are present but lack type information at the parameter level in descriptions. No tool annotations (readOnlyHint/destructiveHint) are present. Results do not document their structure or pagination behavior.
Acknowledge an incident
Create an alert rule for a check
Create a new monitoring check
Trigger immediate execution of a check
Parameter descriptions are incomplete or missing context. E.g., 'check_id' in trigger_check lacks guidance on format or how to obtain it; 'regions' in create_check lacks explanation of valid values or default behavior.
No output/return schemas documented for any tool. LLMs cannot determine what fields to expect or how to chain results to downstream tools. Create_check should document it returns check_id, status, next_run_time, etc.
No tool annotations present. Tools like create_check and create_alert_rule (destructive/state-modifying) should declare destructiveHint=true. Trigger_check should declare idempotentHint=true.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | C | 67 | <=2025-11-25 | v2 |
| 2026-03-09 | D | 51 | - | v1 |
Error responses in code (respondError function) do not provide recovery guidance. HTTP error codes alone do not tell the LLM whether to retry, ask the user, or abort. No mechanism for per-item failures in batch operations.
create_check accepts optional 'regions' array but provides no enum of valid region names. LLMs will hallucinate invalid region IDs. Should constrain to known regions (us-east, eu-west, ap-southeast, etc.) via enum.
No dry-run or confirmation pattern for destructive operations. Create_check, trigger_check (which may fail checks), and create_alert_rule (which may spam notifications) lack safeguards. Agent mistakes could create large numbers of checks/rules unintentionally.
No pagination documented or available on list/discovery tools (ListResources, GetResource). If a customer has thousands of checks, all are returned, consuming unbounded context. No limit parameter or cursor mechanism.