Repeatability
High
The task is structurally identical every time: fixed thresholds, fixed instance count, fixed alert channel, fixed schedule. There is no judgment variation between runs.
Ambiguity Tolerance
High
Success criteria are explicit — 85% disk, 90% memory, Slack alert with instance ID and metric, 15-minute cron interval. A non-human can verify the script meets these criteria by inspection.
Data & Tool Availability
Medium
The agent can write the script without live access, but deployment requires the actual Slack webhook URL, AWS instance IDs or tags, and SSH or SSM access to the instances — none of which the agent has by default.
Error Cost
Low
The script is read-only monitoring; it cannot modify instances or cause outages. A bug means missed alerts or false positives, which is annoying but not catastrophic and is easily corrected.
Human Judgment Required
Low
Threshold values and alert targets are already specified. Decisions about what constitutes a 'stale process' may need minor clarification, but this is a solvable ambiguity with a reasonable default heuristic.