Problem

LLMs often generate invalid PromQL or LogQL queries due to syntax errors. Currently, these errors are only discovered after the query is sent to Prometheus/Loki, which:

• Wastes API calls and context window space
• Creates unnecessary iteration cycles
• Provides unclear error messages
• Slows down LLM query generation workflows

Proposed Solution

Add query validation using the official parsers from Prometheus and Loki before executing queries.

Implementation

Add validation functions using production-grade parsers:

• ValidatePromQL(): Uses github.com/prometheus/prometheus/promql/parser
• ValidateLogQL(): Uses github.com/grafana/loki/v3/pkg/logql/syntax

Integrate inline validation into existing tools:

• query_prometheus: Validate PromQL before execution
• query_loki_logs: Validate LogQL before execution

Dependencies

• Prometheus parser: Already available in github.com/prometheus/prometheus
• Loki parser: Add github.com/grafana/loki/v3/pkg/logql/syntax

Benefits

1. For LLMs: Reduced context usage and faster iteration
2. For users: Immediate feedback on query syntax
3. For API: Fewer invalid requests
4. For debugging: Clear, specific error messages

Testing

Comprehensive unit tests covering valid queries, invalid syntax, and edge cases.

Alternative Considered

Creating separate validate_promql and validate_logql tools was considered, but inline validation is simpler and provides immediate feedback without additional tool calls.