Three read-only tools with basic schemas and descriptions. All tools have input schemas with typed parameters and descriptions, meeting minimum requirements. However, descriptions are terse (50-80 chars), lacking context on WHEN to use each tool or HOW results chain together. No output schemas documented. Parameters lack constraints (enums, ranges, formats). Error handling and recovery guidance absent. Tools follow verb_noun naming but lack composition hints (e.g., which tool to call first, how results feed downstream). No tool annotations (readOnlyHint present in metadata but not in schema). Baseline: avg tool description ~194 chars; these average ~65 chars.

• Expand tool descriptions to 100-150 chars. Add WHEN context: 'Use to find logs matching a time range, level, or route. Call this first to identify the incident window, then use query_metrics to correlate with resource spikes.'
• Document output schemas explicitly. E.g., 'Returns {rows: [{id: int, timestamp: string, level: string, message: string, route: string}], total: int}'. Include fields downstream tools need (e.g., timestamp for sorting, id for filtering).
• Add parameter constraints. 'since' → 'ISO 8601 timestamp (e.g., 2024-01-15T10:30:00Z)'. 'level' → enum: [INFO, WARN, ERROR, DEBUG]. 'limit' → integer, min 1, max 1000, default 100.
• Add composition hints. E.g., query_logs description: 'Returns incident logs. Combine with query_metrics to correlate timing. Use get_recent_commits to identify deploy-related incidents.'
• Add readOnlyHint to all tool schemas: 'readOnlyHint': true. This signals to LLMs that these tools are safe to call without confirmation.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry