# Write Bats Test

> Add or extend a bats-core test for the shell scripts, following the tests/ convention. Use when changing setup.sh or context.sh behavior.

- Skill: `dailybothq/write-bats-test` (Agent Skill)
- Install (CLI): `npx skillmds@latest add dailybothq/write-bats-test`
- Raw SKILL.md: https://api.skillmd.com/api/skills/dailybothq/write-bats-test/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: dailybothq (https://skillmd.com/u/dailybothq)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/dailybothq/write-bats-test

---


# Write Bats Test

## Goal

Cover new or changed shell behavior with a bats-core test that matches the
existing convention in `tests/`.

## When to use

When `setup.sh` or `skills/deepworkplan/shared/context.sh` gains or changes
behavior (a new flag, a new detection branch, a new output field).

## Steps

1. Read the existing tests to match style and helpers:
   `tests/setup-sh.bats`, `tests/context-sh.bats`.
2. Add `@test` cases for the new behavior. Use a throwaway `HOME`/temp dir for
   any filesystem effect so the suite never touches the developer's `~/.claude`.
3. Keep assertions on observable contract: emitted JSON fields from
   `context.sh`, created symlink names from `setup.sh`, exit codes.
4. Run `bats tests/` and iterate until green.

## Validation

`bats tests/` passes, including the new cases, and the suite leaves no artifacts
outside its temp dirs.

## Notes

Tests must run on bash 3.2. Prefer testing the **public contract** (flags,
symlink names, JSON keys) over internal implementation so refactors don't churn
the suite.