@codewizard-dt/bootstrap 2.18.0 → 2.19.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -178,6 +178,16 @@ three is the whole point of the subsystem.
178
178
  **`confirm`**, inside its own `auto | confirm | never` grammar. It has
179
179
  deliberately no separate `ask` value, and offering one is a hard failure —
180
180
  `--set gitCommit.versionBump --value ask` exits 1.
181
+ - **The commit-subject prefix follows the bump**, for that same key. `/git-commit`
182
+ writes `[patch]`/`[minor]`/`[major]` **if and only if** it actually bumped a
183
+ version: written under `auto` and an approved `confirm`, omitted under `never`
184
+ and a declined `confirm`. An earlier contract wrote it unconditionally so that
185
+ release tooling "kept working"; that was backwards. The prefix is a claim about
186
+ what the commit did and tooling acts on it, so a `[minor]` on a commit that
187
+ bumped nothing publishes a release with no version change behind it. Pinned by
188
+ `test/bootstrap-prefs.test.js`, across the skill, the schema `detail`, and the
189
+ installer prompt — a user who picks `never` from the prompt's own description
190
+ and then finds a prefix on their commit was mis-sold.
181
191
 
182
192
  ### The installed layout is what makes skill keys readable
183
193
 
@@ -394,7 +404,7 @@ above; `gitignore.section.*`'s deliberate one-value grammar is in
394
404
  | `gitCommit.versionBump` | either | skill | `auto \| confirm \| never` | `auto` | `install-global.sh` | How `/git-commit` handles the version bump before committing |
395
405
  | `gitCommit.autoPush` | either | skill | `true \| false \| ask` | `false` | `install-global.sh` | Whether `/git-commit` pushes after a successful commit |
396
406
  | `research.persistToRaw` | either | skill | `true \| false \| ask` | `true` | `install-global.sh` | Whether `/research` writes its report and sources to `raw/research/` |
397
- | `uatGenerate.promoteTests` | either | skill | `true \| false \| ask` | `true` | `install-global.sh` | Whether `/uat-generate` promotes UAT checks into repeatable tests under `test/` |
407
+ | `uatGenerate.promoteTests` | either | skill | `sibling \| never \| dedicated` | `dedicated` | `install-global.sh` | Where `/uat-generate` writes the unit tests it promotes out of UAT cases |
398
408
  | `gitignore.offerSectionUpdates` | either | skill | `true \| false \| ask` | `true` | `install-global.sh` | Master gate for the `.gitignore` template section review pass |
399
409
 
400
410
  These five are the entire `scope: either` population and the entire
@@ -214,7 +214,7 @@ else
214
214
  echo ""
215
215
  echo " Should /git-commit bump the version in package.json (and every other manifest) before committing?"
216
216
  echo " Say 'never' for apps and private repos; 'auto' for published packages. The [patch]/[minor]/[major]"
217
- echo " subject prefix is written either way, so release tooling keeps working."
217
+ echo " subject prefix follows the bump — written when one happens, omitted when it doesn't."
218
218
  VERSION_BUMP_ANSWER="$(prompt_letter_choice skip " [a]uto / [c]onfirm each time / [n]ever / [s]kip for now: " auto confirm never skip)"
219
219
  case "$VERSION_BUMP_ANSWER" in
220
220
  auto|confirm|never)
@@ -232,8 +232,49 @@ else
232
232
  "Should /git-commit push the current branch after committing? (Default no — this turns a local action into a published one. It never creates a branch either way.)"
233
233
  settle_skill_pref research.persistToRaw \
234
234
  "Should /research save its report and sources to raw/research/ by default? (Findings always appear in the reply; this governs the file write only, and you can still decline any single report.)"
235
- settle_skill_pref uatGenerate.promoteTests \
236
- "Should /uat-generate write repeatable assertions into test/ automatically, alongside the UAT file?"
235
+ # uatGenerate.promoteTests is the second key that cannot go through
236
+ # settle_skill_pref: its grammar is sibling|never|dedicated (plus the
237
+ # parameterized dedicated:<path>), not true|false|ask.
238
+ #
239
+ # THE OPTION NAMED HERE IS `beside`, THE VALUE STORED IS `sibling`.
240
+ # prompt_letter_choice matches on FIRST LETTER and takes the first listed name
241
+ # that matches, so offering `sibling` alongside `skip` would silently make one
242
+ # of the two unreachable — whichever came second could never be typed. `beside`
243
+ # collides with nothing and reads better in the prompt; the case below maps it
244
+ # back to the schema's value.
245
+ if ! prefs_stored_global uatGenerate.promoteTests; then
246
+ echo ""
247
+ echo " Where should /uat-generate write the unit tests it promotes out of UAT cases?"
248
+ echo " A dedicated folder keeps the suite in one place; beside puts src/parse.ts's test at"
249
+ echo " src/parse.test.ts. A project's existing layout always wins over this answer."
250
+ TEST_LOCATION_ANSWER="$(prompt_letter_choice skip \
251
+ " [d]edicated folder / [b]eside each file / [n]ever write tests / [s]kip for now: " \
252
+ dedicated beside never skip)"
253
+ case "$TEST_LOCATION_ANSWER" in
254
+ dedicated)
255
+ # No follow-up asking WHICH folder. `values` is a closed enumeration of
256
+ # literal tokens, and that closedness is what guarantees no preference key
257
+ # can hold an arbitrary string (see the schema's own no-secrets rule) — a
258
+ # stored path would be exactly that. The skill resolves the directory from
259
+ # the repo instead: an existing test directory, else the language default.
260
+ prefs_set uatGenerate.promoteTests --global dedicated
261
+ echo " uatGenerate.promoteTests = dedicated"
262
+ ;;
263
+ beside)
264
+ prefs_set uatGenerate.promoteTests --global sibling
265
+ echo " uatGenerate.promoteTests = sibling"
266
+ ;;
267
+ never)
268
+ prefs_set uatGenerate.promoteTests --global never
269
+ echo " uatGenerate.promoteTests = never"
270
+ ;;
271
+ *)
272
+ echo " uatGenerate.promoteTests left unanswered — today's behavior (a dedicated test folder) is unchanged, and you will be asked again next sync."
273
+ ;;
274
+ esac
275
+ PREFS_ASKED=$(( PREFS_ASKED + 1 ))
276
+ fi
277
+
237
278
  settle_skill_pref gitignore.offerSectionUpdates \
238
279
  "Should setup/update offer .gitignore template section updates? (No stops the opening question entirely. The .git/info/exclude block is separate and is unaffected.)"
239
280
 
@@ -131,7 +131,7 @@
131
131
  "scope": "either",
132
132
  "consumer": "skill",
133
133
  "summary": "How /git-commit handles the version bump before committing",
134
- "detail": "'auto' is today's behavior: the skill picks a bump level and applies it to every version manifest it finds, without confirmation. 'confirm' prints the suggested bump and asks before editing any manifest — 'confirm' IS this key's ask state, so there is deliberately no separate 'ask' value in the grammar. 'never' touches no version file at all but STILL prefixes the commit subject with [patch], [minor], or [major], because downstream release tooling reads that prefix. This is a choice among three behaviors, not an add/skip gate, so the declines-only rule does not apply — the answer is recorded in all three directions.",
134
+ "detail": "'auto' is today's behavior: the skill picks a bump level and applies it to every version manifest it finds, without confirmation. 'confirm' prints the suggested bump and asks before editing any manifest — 'confirm' IS this key's ask state, so there is deliberately no separate 'ask' value in the grammar. 'never' touches no version file at all. The [patch]/[minor]/[major] commit-subject prefix is written IF AND ONLY IF a version was actually bumped: written under 'auto' and under an approved 'confirm', omitted entirely under 'never' and under a declined 'confirm'. That coupling is the point — release tooling reads the prefix as a claim about what the commit did, so a prefix on a commit that bumped nothing is a false statement the tooling would act on, not a harmless label. This is a choice among three behaviors, not an add/skip gate, so the declines-only rule does not apply — the answer is recorded in all three directions.",
135
135
  "values": "auto | confirm | never",
136
136
  "default": "auto",
137
137
  "askedBy": "install-global.sh"
@@ -157,10 +157,10 @@
157
157
  "uatGenerate.promoteTests": {
158
158
  "scope": "either",
159
159
  "consumer": "skill",
160
- "summary": "Whether /uat-generate promotes UAT checks into repeatable tests under test/",
161
- "detail": "true is today's behavior: alongside the UAT file in wiki/work/uat/, the skill writes repeatable assertions into the project's test/ directory so the checks survive as regression coverage. false generates the UAT file only and adds nothing to test/. 'ask' prompts per invocation. This affects what lands on disk, not how the UAT itself is judged.",
162
- "values": "true | false | ask",
163
- "default": true,
160
+ "summary": "Where /uat-generate writes the unit tests it promotes out of UAT cases",
161
+ "detail": "The question is WHERE promoted tests land, not whether they exist. 'dedicated' is today's behavior: they go in the project's dedicated test directory, resolved by looking at the repo first (an existing test directory) and falling back to the language default — test/ for JS/TS, tests/ for Python, spec/ where that is the local idiom. The directory is detected rather than stored, because this grammar is a closed enumeration of literal tokens and a stored path would be free text. 'sibling' puts each test beside the file under test using the language's own convention (src/parse.ts -> src/parse.test.ts). 'never' writes nothing: the skill still runs its promotability analysis and records the verdict on each UAT case, it just creates no file. An existing test file that already covers the unit is extended in place under every value, and a project's visible layout wins over a stored answer that contradicts it — so this key is a default for new work, not an override of the repo. It affects what lands on disk, not how the UAT itself is judged. Legacy true/false/ask values from the key's earlier grammar are mapped by the skill: true -> dedicated, false -> never, ask -> ask once per run.",
162
+ "values": "sibling | never | dedicated",
163
+ "default": "dedicated",
164
164
  "askedBy": "install-global.sh"
165
165
  },
166
166
  "gitignore.offerSectionUpdates": {
@@ -72,12 +72,28 @@ Print the summary. What happens next is decided by `gitCommit.versionBump` from
72
72
 
73
73
  Its grammar is `auto | confirm | never` — there is deliberately **no `ask` value**, because `confirm` *is* this key's ask state. Do not look for one.
74
74
 
75
- | Value | What to do |
76
- |-------|------------|
77
- | `auto` | Today's behavior. Detect every manifest and apply the bump to all of them, no confirmation. |
78
- | `confirm` | Print the suggested bump **and the full list of manifest files that would change**, then ask once with `AskUserQuestion` before editing anything. Approved → bump exactly as `auto` does. Declined → edit no manifest, and still commit with the subject prefix. Ask **every run** and **never persist the answer** — the standing choice is "keep asking me", so recording a reply would destroy it. |
79
- | `never` | Touch **no** version file at all. Skip straight to Step 5. The `[patch]`/`[minor]`/`[major]` subject prefix is still written, because downstream release tooling reads that prefix — dropping it would break releases, not just skip a bump. Say so once in the final report: `Version files untouched (gitCommit.versionBump=never); the subject prefix is still set. Change with /bootstrap-config.` |
80
- | `unset` | Behave exactly as `auto` — that is today's behavior and an unanswered key must never change it. Note once in the final report: `gitCommit.versionBump is unanswered — /git-commit bumps every manifest by default. Set it with: node ~/.claude/bootstrap-prefs.js --set gitCommit.versionBump --value auto --global` (or `confirm` / `never`). |
75
+ | Value | Manifests | Subject prefix |
76
+ |-------|-----------|----------------|
77
+ | `auto` | Today's behavior. Detect every manifest and apply the bump to all of them, no confirmation. | **written** |
78
+ | `confirm` — approved | Bump exactly as `auto` does. | **written** |
79
+ | `confirm` — declined | Edit no manifest. | **omitted** |
80
+ | `never` | Touch **no** version file at all. Skip straight to Step 5. | **omitted** |
81
+ | `unset` | Behave exactly as `auto` — that is today's behavior and an unanswered key must never change it. | **written** |
82
+
83
+ `confirm` prints the suggested bump **and the full list of manifest files that would change**, then asks once with `AskUserQuestion` before editing anything. Ask **every run** and **never persist the answer** — the standing choice is "keep asking me", so recording a reply would destroy it.
84
+
85
+ ### ⛔ The prefix and the bump travel together
86
+
87
+ **Write `[patch]`/`[minor]`/`[major]` if and only if a version was actually bumped in this run.** The prefix is a claim about what the commit did, and release tooling reads it as one. A `[minor]` prefix on a commit that bumped nothing is not a harmless label — it is a false statement that tooling will act on, tagging or publishing a release that has no version change behind it. An unprefixed subject is merely uninformative; a lying prefix is actively wrong, and it is worse the more automation is downstream.
88
+
89
+ So when no bump happens, commit a **bare subject with no bracket prefix at all**. Do not substitute `[none]`, `[skip]`, `[no-bump]`, or any other placeholder — an unrecognized token is its own kind of noise, and tooling that pattern-matches the three real values will ignore a bare subject correctly.
90
+
91
+ The one case where nothing was bumped for a reason unrelated to this preference — a repo with **no manifest at all** (Detection below finds nothing) — follows the same rule: nothing was bumped, so no prefix.
92
+
93
+ Report lines, each at most once per run:
94
+ - `never` → `Version files untouched and no subject prefix written (gitCommit.versionBump=never). Change with /bootstrap-config.`
95
+ - `confirm`-declined → `Bump declined — no manifest edited and no subject prefix written.`
96
+ - `unset` → `gitCommit.versionBump is unanswered — /git-commit bumps every manifest by default. Set it with: node ~/.claude/bootstrap-prefs.js --set gitCommit.versionBump --value auto --global` (or `confirm` / `never`).
81
97
 
82
98
  Anything else stored under this key is a value from a newer bootstrap that this skill has no branch for: treat it as `unset`, and say which value you did not recognize.
83
99
 
@@ -128,7 +144,7 @@ Edit **every file** that contains a version using the Edit tool — do not stop
128
144
  - **ALWAYS use the `git-commit` bash alias.** Never use `git add` or `git commit` directly.
129
145
  - Always commit ALL files unless they are in `.gitignore`
130
146
  - Consider an appropriate commit message, let's call it `$message`
131
- - Prefix the subject with the bump type in brackets: `[patch]`, `[minor]`, or `[major]`
147
+ - Prefix the subject with the bump type in brackets — `[patch]`, `[minor]`, or `[major]` — **only if Step 4 actually bumped a version**. If it did not (`never`, `confirm`-declined, or no manifest in the repo), the subject starts with the description and carries no bracket prefix
132
148
  - Run: `git-commit "$message"`
133
149
 
134
150
  ### ⛔ No agent attribution — the user is the sole author
@@ -143,13 +159,14 @@ The commit must be attributed to the user's git identity **only**. Never add the
143
159
 
144
160
  The commit message MUST be a **single-line, single-quoted-argument string**. Subject only, no body. This is a hard rule because anything else triggers a Bash approval prompt and slows you down.
145
161
 
146
- **Required form:**
162
+ **Required form** — bumped, and not bumped:
147
163
 
148
164
  ```bash
149
165
  git-commit "[patch] Single-line subject describing the change"
166
+ git-commit "Single-line subject describing the change"
150
167
  ```
151
168
 
152
- That's it. One pair of double quotes. One line. One argument.
169
+ That's it. One pair of double quotes. One line. One argument. The only difference between the two forms is whether Step 4 bumped a version.
153
170
 
154
171
  **❌ Forbidden — every one of these patterns triggers an approval prompt:**
155
172
 
@@ -178,9 +195,13 @@ git-commit "subject" "-m" "body"
178
195
 
179
196
  # WRONG — piping or chaining
180
197
  echo "subject" | git-commit -F -
198
+
199
+ # WRONG — a placeholder standing in for a bump that did not happen
200
+ git-commit "[none] Fix the login redirect"
201
+ git-commit "[no-bump] Fix the login redirect"
181
202
  ```
182
203
 
183
- **✅ Correct — always:**
204
+ **✅ Correct — a version was bumped:**
184
205
 
185
206
  ```bash
186
207
  git-commit "[patch] Fix UAT verdict logic for edge-case test skips"
@@ -188,14 +209,23 @@ git-commit "[minor] Add /task-audit skill with dependency graph and wave output"
188
209
  git-commit "[major] Rename all skills to noun-first convention, drop legacy verb-first aliases"
189
210
  ```
190
211
 
212
+ **✅ Correct — no version was bumped** (`never`, `confirm`-declined, or no manifest in the repo):
213
+
214
+ ```bash
215
+ git-commit "Fix UAT verdict logic for edge-case test skips"
216
+ git-commit "Add /task-audit skill with dependency graph and wave output"
217
+ ```
218
+
191
219
  ### Writing a good single-line subject
192
220
 
193
221
  If the change feels too complex to summarize in one line, **make the subject more descriptive** — do not reach for a multi-line body. Aim for ~70–100 characters but go longer if it adds real information. Lead with the *what* and the *why* on the same line, separated by a colon when helpful:
194
222
 
195
223
  - ✅ `[minor] Require research for every test type in /uat-generate with checkpoint gate`
196
224
  - ✅ `[patch] Allow standard tools for markdown editing; ban bash exploration commands`
197
- - ❌ `Update commands` (too vague, missing bump prefix)
225
+ - ✅ `Allow standard tools for markdown editing; ban bash exploration commands` (no bump this run — the subject still carries its weight)
226
+ - ❌ `Update commands` (too vague — and vague is vague with or without a prefix)
198
227
  - ❌ `[patch] Fix bug` (too vague)
228
+ - ❌ `[patch] Fix the login redirect` **when Step 4 bumped nothing** (the prefix is false)
199
229
 
200
230
  Detailed reasoning, before/after examples, and rationale belong in **PR descriptions**, not in commit message bodies. The commit subject is the index entry; the PR is the encyclopedia.
201
231
 
@@ -89,7 +89,7 @@ Do not proceed to Step 3 until you can answer yes to all:
89
89
 
90
90
  ##### Preference gate
91
91
 
92
- This gate governs **only whether files are written under `test/`**. It never affects the UAT file itself or how UAT verdicts are judged.
92
+ This gate governs **only where unit test files are written** (and whether they are written at all). It never affects the UAT file itself or how UAT verdicts are judged.
93
93
 
94
94
  Read the preference **once per invocation** (not per case), before Step 2.5 does any other work:
95
95
 
@@ -97,12 +97,25 @@ Read the preference **once per invocation** (not per case), before Step 2.5 does
97
97
  node ~/.claude/bootstrap-prefs.js --get uatGenerate.promoteTests --project . 2>/dev/null || echo unset
98
98
  ```
99
99
 
100
+ The key answers **where** promoted tests go, not whether they exist. Its grammar is the closed enumeration `sibling | never | dedicated`.
101
+
100
102
  | Value | Behaviour |
101
103
  |-------|-----------|
102
- | `true` | Today's behaviour — run the promotion checkpoint and create/update the unit test files. |
103
- | `false` | Do **not** create or update anything under `test/`. Still run the promotability *analysis* and still record it per case, as `- **Repeatable Unit Test**: Skipped by preference (uatGenerate.promoteTests=false)`. Say once in the completion report (Step 7): "No test/ files were written (uatGenerate.promoteTests=false). Change with /bootstrap-config." |
104
- | `ask` | Ask the user **once per invocation** (AskUserQuestion), before Step 2.5 does any work, whether to promote tests this run. Honour the answer for this run only — never write it back to the store. |
105
- | `unset` | Promote (today's behaviour) and note once: "uatGenerate.promoteTests is unanswered — /uat-generate promotes tests by default. Set it with: node ~/.claude/bootstrap-prefs.js --set uatGenerate.promoteTests --value <true\|false\|ask> --global" |
104
+ | `dedicated` | Write promoted tests into the project's dedicated test directory, **resolved from the repo** — an existing test directory if there is one, else the language default (`test/` for JS/TS, `tests/` for Python, `spec/` where that is the local idiom). The directory is detected, never stored: `values` is a closed list of literal tokens, and a stored path would be free text. |
105
+ | `sibling` | Write each test **beside the file under test**, using the language's own convention — `src/parse.ts` → `src/parse.test.ts`, `src/parse.js` → `src/parse.test.js`, `parse.py` → `test_parse.py`, `parse.go` → `parse_test.go`. Never invent a suffix the language's runner does not already discover. |
106
+ | `never` | Create or update **nothing** under any test path. Still run the promotability *analysis* and still record it per case, as `- **Repeatable Unit Test**: Skipped by preference (uatGenerate.promoteTests=never)`. Say once in the completion report (Step 7): "No test files were written (uatGenerate.promoteTests=never). Change with /bootstrap-config." |
107
+ | `unset` | Behave as `dedicated` — today's behaviour. Note once: "uatGenerate.promoteTests is unanswered — /uat-generate writes promoted tests to a dedicated folder by default. Set it with: node ~/.claude/bootstrap-prefs.js --set uatGenerate.promoteTests --value <sibling\|never\|dedicated> --global" |
108
+ | `true` / `false` / `ask` | **Legacy values** from the key's earlier true/false grammar. Map them and carry on — do not error: `true` → `dedicated`, `false` → `never`, `ask` → ask the user once per invocation (AskUserQuestion) which of the three locations to use, honouring the answer for this run only and never writing it back. Mention once that the stored value is from an older grammar and `/bootstrap-config` can update it. |
109
+
110
+ ### The project's layout outranks the stored answer
111
+
112
+ **A stored location is a default for new work, not an instruction to fight the repo.** Before writing anything, look at where tests already live:
113
+
114
+ - An existing test file that already covers the unit under test is **extended in place** under every value except `never`. That is neither "dedicated" nor "sibling" — it is the local pattern, and it always wins.
115
+ - If the project visibly uses one layout and the stored preference says the other (preference says `sibling`, the repo has a populated `test/` with no siblings anywhere), **follow the repo** and say so once in Step 7: "Followed the project's existing `test/` layout over uatGenerate.promoteTests=sibling. Set it per-project with /bootstrap-config." Scattering siblings through a repo that centralises its suite makes the suite worse, and this key is `scope: either` precisely so a project can hold its own answer.
116
+ - Only when the project has no discernible convention does the stored value decide on its own.
117
+
118
+ `never` is the one value the repo cannot overrule — "write no test files" is a directive, not a layout hint.
106
119
 
107
120
  For every planned UAT case, decide whether any part of it can be captured as a repeatable unit test with low setup cost.
108
121
 
@@ -117,13 +130,13 @@ A UAT case is **not** unit-test promotable when it primarily verifies:
117
130
  - Real HTTP service wiring, authentication middleware, deployment behavior, shell environment setup, database migrations, or file permissions.
118
131
  - Third-party service behavior or integration between multiple independently deployed systems.
119
132
 
120
- Before writing the UAT file, inspect the existing test layout and runner:
121
- - Find nearby test files for the changed code.
133
+ Before writing the UAT file, inspect the existing test layout and runner — this is also what resolves the location gate above:
134
+ - Find nearby test files for the changed code, and note **where** the project keeps them (a dedicated directory, siblings, or both).
122
135
  - Identify the test framework, naming convention, fixtures, mocks, and command used by the repo.
123
- - Prefer adding focused assertions to an existing nearby test file when that is the local pattern.
124
- - Otherwise create the smallest conventional new test file beside the unit under test.
136
+ - Prefer adding focused assertions to an existing nearby test file when that is the local pattern — this wins over the stored preference in every case except `never`.
137
+ - Otherwise create the smallest conventional new test file **at the location the gate resolved to**, following the repo's visible layout when it contradicts the stored value.
125
138
 
126
- If a case is promotable **and the preference gate allows promotion**, you **must** create or update the unit test file. Do not merely list it as a recommendation. If you cannot create it because the test framework is absent or the harness would require new infrastructure, record that as a gap with the concrete blocker. When the gate suppresses promotion, do the analysis and record `Skipped by preference` — write nothing under `test/`.
139
+ If a case is promotable **and the gate resolved to a location** (anything but `never`), you **must** create or update the unit test file at that location. Do not merely list it as a recommendation. If you cannot create it because the test framework is absent or the harness would require new infrastructure, record that as a gap with the concrete blocker. When the gate resolved to `never`, do the analysis and record `Skipped by preference` — write no test file anywhere.
127
140
 
128
141
  ### Step 3: Generate UAT test cases
129
142
 
@@ -210,7 +223,7 @@ implements::[[TASK-NNN]]
210
223
  - Add `- **Repeatable Unit Test**: Created: \`path/to/test-file\`` when you created or updated a unit test for the behavior.
211
224
  - Add `- **Repeatable Unit Test**: Not applicable: <short reason>` when the case cannot reasonably become a unit test.
212
225
  - Add `- **Repeatable Unit Test**: Blocked: <short reason>` only when it was promotable in principle but the repo lacks the needed test harness or fixtures.
213
- - Add `- **Repeatable Unit Test**: Skipped by preference (uatGenerate.promoteTests=false)` when the Step 2.5 preference gate suppressed promotion for this run. Record the promotability analysis as usual; only the file write is skipped.
226
+ - Add `- **Repeatable Unit Test**: Skipped by preference (uatGenerate.promoteTests=never)` when the Step 2.5 preference gate suppressed promotion for this run. Record the promotability analysis as usual; only the file write is skipped.
214
227
 
215
228
  **Unit test command** (required whenever, and only when, the line above says `Created:`):
216
229
  - Add `- **Unit Test Command**: \`<cmd>\`` on the next line, holding the **file-scoped** command that runs exactly that test file — e.g. `node --test test/foo.test.js`, `npx vitest run test/foo.test.ts`, `pytest tests/test_foo.py`, `go test ./pkg/foo/`.
@@ -228,7 +241,7 @@ Auth-Role: user
228
241
 
229
242
  ### Step 4: Write repeatable unit tests, the UAT file, and cross-reference
230
243
 
231
- 1. **If the Step 2.5 preference gate allows promotion**, create or update the actual unit test file for each promotable case from Step 2.5, before writing the UAT file. If the gate suppressed promotion, skip this step entirely — write nothing under `test/` — and record each promotable case as `Skipped by preference` instead.
244
+ 1. **If the Step 2.5 gate resolved to a location**, create or update the actual unit test file for each promotable case from Step 2.5 **at that location**, before writing the UAT file. If it resolved to `never`, skip this step entirely — write no test file anywhere — and record each promotable case as `Skipped by preference` instead.
232
245
  - Use existing test framework conventions and nearby examples.
233
246
  - Keep the test focused on the behavior from the UAT case.
234
247
  - Do not add broad snapshots or duplicate end-to-end checks as unit tests.
@@ -236,7 +249,7 @@ Auth-Role: user
236
249
  - If editing code test files, use Serena editing tools.
237
250
  - If creating new config/markdown/test files where Serena is unavailable, use the appropriate project-approved edit/write tool; never shell redirection.
238
251
 
239
- 2. Include the created test path in the corresponding UAT case's `Repeatable Unit Test` metadata, plus the verified file-scoped `Unit Test Command` — or `Skipped by preference (uatGenerate.promoteTests=false)` when the gate suppressed promotion.
252
+ 2. Include the created test path in the corresponding UAT case's `Repeatable Unit Test` metadata, plus the verified file-scoped `Unit Test Command` — or `Skipped by preference (uatGenerate.promoteTests=never)` when the gate suppressed promotion.
240
253
 
241
254
  3. Write `wiki/work/uat/UAT-NNN-slug.md` using the `Write` tool.
242
255
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@codewizard-dt/bootstrap",
3
- "version": "2.18.0",
3
+ "version": "2.19.0",
4
4
  "description": "Bootstrap Claude Code projects with an LLM Wiki, reusable skills, hooks, and MCP setup",
5
5
  "homepage": "https://github.com/codewizard-dt/bootstrap-wiki#readme",
6
6
  "bugs": {