Execution Model — MANDATORY

You MUST use the Agent tool to execute this skill. Do NOT run the analysis in the main conversation — spawn a general-purpose Agent instead. This keeps the main context clean (the script runs ~60s with 50 parallel MiniMax calls and produces large output).

How to invoke

If the user didn't provide a session UUID, find the most recent session first (see "Find Recent Sessions" below), then ask the user which session to analyze.
Spawn an Agent with the full instructions from this skill and the resolved session UUID. The agent prompt should include: the bun command to run, the --shots flag only if the user explicitly requested fewer, and instructions to present the output as-is.
When the Agent returns, present a concise summary of findings to the user — CRITICAL and WARNING items first, then the Priority Action Plan.

Example Agent invocation:

Agent(
  description: "Session blind spot analysis",
  prompt: "Run session blind spot analysis for session <UUID>. Execute: bun run $HOME/eon/cc-skills/plugins/devops-tools/scripts/session-blind-spots.ts <UUID> ... [full command with flags]. Present the output directly — do not summarize or filter. If the command fails, show the error and suggest troubleshooting steps.",
  run_in_background: true
)

Related skills