What problem does it solve? When a program crashes, a board reboots unexpectedly, or a service refuses to start on a D-Robotics RDK device, developers waste time pasting raw logs back and forth. This Skill turns scattered signals in dmesg, journalctl, and /var/crash into a structured evidence list for root-cause analysis. ## Core Features & Use Cases - Structured log collection: Runs a read-only script that counts kernel err-level messages, lists failed systemd units, detects coredumps and /var/crash files, and checks whether the last boot shut down cleanly, outputting JSON. - Guided root-cause workflows: Provides per-symptom playbooks for crashed services, sudden reboots, and user-pasted errors, with handoffs to sibling skills for thermal, memory, and model-inference issues. - Use Case: A user's board rebooted overnight. The agent runs the collector, finds last_boot.clean is false and a thermal error in kernel.recent_errors, then presents an evidence-to-hypothesis-to-verification chain instead of guessing. ## Quick Start Ask the agent to check the board's logs for clues about a crash or unexpected reboot and summarize the evidence.
