What problem does it solve? Answering "why does this code work this way" usually means digging through git blame, PR discussions, tickets, docs, and chat threads by hand. This Skill automates that investigation and returns a cited, confidence-labeled read on the decisions and tradeoffs that shaped the code. ## Core Features & Use Cases - Parallel evidence investigators: Discovers available MCP servers and spawns one investigator per evidence category (source control, issue tracker, long-form docs, team chat, infrastructure observability, error tracking, product analytics), each with its own playbook. - Epistemic synthesis: A synthesizer subagent labels every claim as measured, inferred, or guess, and produces a structured report with competing hypotheses, unknowns, and a per-source coverage map including null results. - Guardrails: Intake checks, explicit HOLD conditions, and cut-first defaults prevent uncited guesses and unjustified skipped sources. - Use Case: Ask why a retry loop with a specific backoff constant exists; the Skill traces the commits, PRs, incident postmortems, and Sentry errors behind it and returns a cited explanation with a Preserve/Change/Avoid/Risk constraint set for planning modifications. ## Quick Start Ask why the timeout handling in src/api/client.ts was implemented this way and request a cited investigation of its design rationale.
