goal-test

1# Write done as a script
2 
3Theory: [The four parts of a loop](https://secondbrainos.dev/#/read/course-2-loop/the-four-parts)
4and the [goal test build page](https://secondbrainos.dev/#/read/track-loop/build-goal-test).
5"Improve the error handling" cannot terminate a loop, because nothing can ever
6say it is finished. The goal must be phrased so that a program — not a person,
7not the model — returns true or false against it.
8 
9## Workflow
10 
111. **Extract the claim.** Ask what the user wants true at the end, then
12   restate it as verifiable facts: which command exits 0, which file exists,
13   which string appears, which number crosses which threshold. If a part
14   cannot be checked by a program, say so and negotiate it down to what can.
152. **Prefer checks that already exist.** The test suite, the linter, the
16   build, the type checker. A goal test that shells out to `make test` is
17   better than one that reimplements it.
183. **Write `goal-test.sh`** (or `.py` if the checks are easier there) in the
19   project root: runs every check, prints one line per check with pass/fail,
20   exits 0 only when all pass. Keep it under ~40 lines; it must run in
21   seconds and be safe to run repeatedly.
224. **Run it now.** It should fail before the work is done — a goal test that
23   passes on the current state is testing nothing. Show the failing output.
245. **Offer the loop.** If the user wants the agent driven until done, wrap it:
25 
26```bash
27#!/usr/bin/env bash
28…