diagnostics_run

Management

Run Diagnostics

Diagnose Scenario MCP problems. Runs a live self-test with the caller's own credentials to identify the failing layer (auth / tenant-scope / api-unreachable / healthy), returns Scenario trace ids, and produces a copy-paste report for Scenario support. Args: - observed_trace_ids: optional real "mcpt_" trace ids seen in the session (preferred over reconstructed ids) - activity_age_minutes / activity_span_minutes: optional relative timing used only to reconstruct candidate trace ids - team_id, project_id: pass the same values used when the problem occurred - response_format: 'json' (default) or 'markdown' Returns: { confirmed_trace_ids, candidate_trace_ids, self_test, auth_type, tenant, endpoint, version } plus a markdown report. Tip: prefer running the /diagnose prompt, which assembles the full report including a tool-call transcript.

read-onlyopen-world

Parameters

Name	Type	Required	Description
team_id	string	—	Team ID. Required if user belongs to multiple teams.
project_id	string	—	Project ID to scope the operation to.
observed_trace_ids	array	—	Real mcpt_ trace ids observed in the session (e.g. from error lines). Preferred over reconstructed ids.
activity_age_minutes	number	—	Roughly how many minutes ago the Scenario activity happened. Used only to reconstruct candidate ids.
activity_span_minutes	number	—	Roughly over how many minutes the Scenario activity spanned.
response_format	enum(json \| markdown)	json	Output format: 'json' for structured data, 'markdown' for human-readable text.