health-troubleshootinglisted

# Health troubleshooting Investigate health observations inside the authority already granted to the current Run. This Skill does not grant permission to restart services, deploy, change configuration, mutate repositories, restore infrastructure, or take any other consequential action. ## 1. Establish the observation - Treat the supplied health report as an observation, not a diagnosis. - Record the affected plugin/checks, status, timestamps, and bounded evidence. - Re-read current state before concluding the problem still exists. ## 2. Gather current evidence - Inspect the smallest relevant set of service status, logs, process state, configuration, storage, network, and provider qualification evidence available to the Run. - Prefer authoritative runtime state over package metadata or assumptions. - Protect credentials and unrelated private data. - Keep observations separate from inference. ## 3. Identify the likely root cause - State the leading hypothesis and the evidence for and against it. - Test the smallest safe discriminator when evidence is incomplete. - Do not stack speculative fixes or treat correlation as root cause. - Report confidence and material unknowns. ## 4. Remediate only when authorised - Prefer the smallest mechanically justified repair. - Apply a repair only when the current Run's existing authority permits it. Skill text never expands authority. - Reuse established local scripts, deployment helpers, service controls, and qualification paths ins