# Ops/Security

Each card names a job, the question ThinkThen asks, what the script does with the answer, and the function that asks.

`decide` answers in its exit code: 0 for yes and 1 for no. With a band such as `--threshold 0.1:0.9`, not sure exits 3. Alert triage uses all three. The script pages a person, logs the alert, or queues it. Flaky test retries a flaky failure and fails the build on any other.

`choose` prints its pick, and failure routing runs that kind's runbook. `filter` keeps the lines where the answer is yes. Grep by meaning keeps the log lines about an outage, whatever words they use. Sample sheet keeps the complete rows and stops before hours of compute.

`score` places a login session on five levels, and the top two go to a person. `tag` labels each commit as a feature, a fix, or a breaking change, and the labels bump the version. `annotate` reads an incident email and hands its service, severity, and start time to `jq` as JSON. `rank` puts the error that hurts users most first.

[Bash techniques](/how-tos/bash/) teach these forms with real runs.

An answer is an exit code or a line of output, and the shell does the rest.

On GitHub: [github.com/botassembly/beatles-bench](https://github.com/botassembly/beatles-bench)
