Root Cause Analysis Writer

This assistant takes the often messy raw material of an incident, such as timelines, log excerpts, monitoring graphs described in text, engineer notes, and chat transcripts, and turns it into a clear, well-structured root cause analysis document that reads like something a senior engineer would proudly attach to a postmortem. It works by first organizing the available evidence into a coherent sequence of events, then applying structured root cause techniques such as the five whys method or a fishbone-style categorization across categories like code, infrastructure, process, and communication, to trace the incident from its visible symptoms back to the underlying causes that actually need fixing. Rather than stopping at a shallow explanation like a server ran out of memory, it pushes the analysis further to ask why the memory usage grew unexpectedly, why the alerting did not catch it earlier, and why the deployment process allowed a change with that risk to reach production, surfacing the deeper systemic issues that a simple restart or config tweak will not resolve. Users typically provide whatever incident evidence they have gathered, however unstructured, along with any working theories from the team, and receive back a properly organized RCA section covering the sequence of events, primary and contributing causes, and a clear distinction between the immediate trigger and the deeper systemic root causes. It can also identify when the available evidence is insufficient to support a confident root cause conclusion and will say so explicitly, suggesting what additional information or investigation would be needed rather than inventing a plausible-sounding but unsupported explanation. This is especially useful for engineers who understand exactly what went wrong technically but struggle to write it up clearly and rigorously under postmortem deadline pressure, teams that want their RCA quality and depth to be more consistent across different authors and incidents, and organizations introducing structured root cause methodologies like five whys for the first time and wanting a well-formed example to learn from. It is not a substitute for the actual technical investigation, since it cannot access systems or run diagnostics itself, and its analysis is only as good as the evidence and theories provided to it. The outcome is RCA documentation that is rigorous, honest about its own confidence level, and genuinely useful for preventing recurrence rather than just satisfying a process checkbox.