rca

Analyze macOS unified logs and session forensics to diagnose system failures.

19|2|Updated Mar 30, 2026
One-click install
npx skills add https://github.com/bjornjee/agent-dashboard --skill rca-bjornjee
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rca
Source: https://github.com/bjornjee/agent-dashboard/tree/main/adapters/claude-code/skills/rca
Command: npx skills add https://github.com/bjornjee/agent-dashboard --skill rca-bjornjee

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you perform root cause analysis for process crashes, server deaths, and unexplained system failures by collecting and validating evidence first, then reasoning from that evidence.

Core Features & Use Cases

  • Evidence-first incident investigation: Scopes the incident, establishes a time window, and enumerates active sessions/processes before any causal reasoning.
  • macOS unified log forensics: Collects process events, AMFI/code-signing failures, termination signals, jetsam/memory pressure, sleep/wake activity, kernel events, and crash reports across the same window.
  • Session forensics and last-command tracing: Extracts Claude-executed Bash tool calls from session JSONL, flags dangerous commands, identifies the final command before the crash, and traces how code paths could cascade.
  • Structured RCA deliverable: Produces an evidence-backed timeline, root cause, contributing factors, ruled-out alternatives, and concrete prevention recommendations.

Quick Start

Run the RCA skill with your incident description (for example, describe which process died, when it started failing, and what you observed), and the skill will guide you through evidence gathering before concluding a root cause.

Frequently Asked Questions about rca

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I perform root cause analysis for a process crash on macOS?

Root cause analysis for a process crash involves collecting evidence from macOS unified logs, session forensics, and crash reports within a specific time window before reasoning about the cause.

How do I trace the last command executed before a server crash in a tmux session?

To trace the last command before a server crash, you can parse session JSONL logs to extract Bash tool calls, flag dangerous commands, and identify the final command executed in the tmux session.

Can I use macOS unified logs to investigate AMFI code-signing failures and memory pressure?

Yes, macOS unified logs provide the necessary forensics to investigate AMFI code-signing failures, jetsam memory pressure events, termination signals, and kernel events to determine why a process died.

What is the best way to structure an incident timeline for unexplained system failures?

The best way to structure an incident timeline for unexplained system failures is to use phase-gated evidence collection, creating a timestamped event chain that includes ruled-out alternatives and contributing factors.

Why does my Claude Code agent execution terminate unexpectedly and how do I diagnose it?

Diagnosing unexpected Claude Code agent execution terminations requires extracting executed Bash commands from session logs and tracing code path cascades to identify the crash trigger.

Does evidence-first incident investigation require crash reports to determine the root cause?

Evidence-first incident investigation uses crash reports alongside process events and session forensics to validate the root cause, but it also produces ruled-out alternatives if crash reports are insufficient.