debug-conductor

What problem does it solve?

Debug Conductor orchestrates bounded, conductor-led investigations for complex distributed-system incidents that cross service boundaries and resist single-agent diagnosis, ensuring investigations produce auditable incident packets and prioritized next steps.

Core Features & Use Cases

• Fork decision & rubric: Applies a fork-readiness rubric to decide between serial and parallel investigations and records score and rationale.
• Parallel investigator orchestration: Selects and dispatches up to four specialized investigators with scoped mandates, token and wall-clock budgets, and non-communicating constraints.
• Evidence validation & adjudication: Validates, deduplicates, and monitors branch evidence, then forwards results to an evidence-adjudicator that produces a root-cause verdict with confidence and alternatives.
• Bounded constraints & safety: Enforces branch limits, budget caps, confidence gates, and duplicate merging to prevent cost explosions and circular work.
• Use cases: Multi-service outages, stalled debugging after serial triage, incidents touching telephony/LLM/state/infra domains, and time-sensitive production incidents.

Quick Start

Invoke the conductor to investigate payment failures across checkout and fulfillment services since 2pm UTC with logs, traces, and failed transaction records attached.