Log Debugging MCP Reference

Overview

This page covers the log debugging MCP tools for direct integration with the gloria.dev MCP server. It stays at the raw tool contract only. For the higher-level workflow, see Debugging and the debugging-production-errors skill reference.

The log debugging surface reads live data from the connected provider, returns structured records, and reports capability status before any query.

Runtime behavior

  • get_log_provider_status reports whether a provider is configured and which log debugging capabilities it supports.
  • query_logs returns structured attributes on each record, accepts one exact attribute key/value filter, and can stop early with a truncation signal.
  • A failure on one connection does not hide successful results from other connections.

get_log_provider_status

Purpose: Check whether the connected log provider is ready and which capabilities it exposes.

Returns: Provider status, configuration state, and capability flags.

Behavior:

  • Use it first to confirm that log debugging is available.
  • Use the capability flags to decide whether structured querying and the other log tools are supported.

get_log_stats

Purpose: Summarize log activity across a scope and time window.

Returns: Summary counts and trend data that help spot spikes or quiet periods.

Behavior:

  • Use it to narrow the incident window before reviewing individual records.
  • Use it before query_logs when the caller needs a smaller time range.

list_error_groups

Purpose: Group repeated failures into error clusters.

Returns: Grouped error summaries plus context for each group.

Behavior:

  • Includes first-seen information and service version details.
  • Helps rank failures before reviewing individual records.

query_logs

Purpose: Return the matching log records for a scope and time window.

Returns: Matching records, structured attributes attached to each record, and a truncation signal when the query stops early.

Behavior:

  • Supports one exact attribute key/value filter.
  • Pushes supported structured filters down to the provider so the search narrows at the source.
  • Keeps successful records from other connections visible even when one connection fails.
  • Returns partial results when the query reaches its limit or ends early.

See also