@monte3l/groundwork 1.0.4 → 1.0.5

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.
Files changed (33) hide show
  1. package/README.md +54 -12
  2. package/dist/harness/rules.js +1 -1
  3. package/dist/main.js +5 -15
  4. package/dist/merge-json.d.ts +3 -1
  5. package/dist/merge-json.js +1 -1
  6. package/package.json +1 -1
  7. package/plugin/skills/customize/step-3-round-1.md +25 -17
  8. package/plugin/src/domain-map.ts +1 -1
  9. package/plugin/src/pack-map.ts +125 -44
  10. package/plugin/src/plugin-map.ts +1 -1
  11. package/templates/core/.claude/hooks/inject-decision-gate.mjs +4 -0
  12. package/templates/core/.claude/rules/agent-dispatch.md +46 -0
  13. package/templates/core/CLAUDE.md +41 -60
  14. package/templates/core/bin/check-exports.mjs +3 -1
  15. package/templates/core/bin/lib/harness-rules.mjs +1 -1
  16. package/templates/core/eslint.config.js +2 -0
  17. package/templates/core/vitest.config.ts +4 -3
  18. package/templates/packs/README.md +6 -3
  19. package/templates/packs/{github → claude-github}/files/.github/workflows/claude-pr-review.yml +1 -1
  20. package/templates/packs/{github → claude-github}/files/.github/workflows/claude.yml +1 -1
  21. package/templates/packs/claude-github/pack.json +19 -0
  22. package/templates/packs/github/pack.json +3 -3
  23. package/templates/packs/harness-extras/pack.json +3 -15
  24. package/templates/packs/licensing/pack.json +31 -0
  25. package/templates/packs/publishing/pack.json +3 -14
  26. package/templates/packs/statusline/pack.json +31 -0
  27. package/templates/packs/supply-chain/pack.json +1 -1
  28. package/templates/packs/worktrees/pack.json +2 -1
  29. /package/templates/packs/{publishing → licensing}/files/REUSE.toml +0 -0
  30. /package/templates/packs/{publishing → licensing}/files/bin/check-license-headers.mjs +0 -0
  31. /package/templates/packs/{harness-extras → statusline}/files/.claude/hooks/statusline-layout.mjs +0 -0
  32. /package/templates/packs/{harness-extras → statusline}/files/.claude/hooks/statusline.mjs +0 -0
  33. /package/templates/packs/{harness-extras → statusline}/files/.claude/hooks/subagent-statusline.mjs +0 -0
package/README.md CHANGED
@@ -44,14 +44,53 @@ Read this project's rendered docs site at
44
44
 
45
45
  ## Quickstart
46
46
 
47
- 1. Install [Node 24+](https://nodejs.org/) and [pnpm](https://pnpm.io/installation)
48
- (see ["Requirements"](#requirements) below for exact versions).
49
- 2. Run `npx @monte3l/groundwork my-app` to bootstrap a new project
50
- named `my-app`.
51
- 3. `cd my-app`.
52
- 4. Open the `my-app` folder in Claude Code.
53
- 5. Run `/customize` -- it interviews you and tailors the baseline to your
54
- project.
47
+ Install [Node 24+](https://nodejs.org/), [pnpm](https://pnpm.io/installation)
48
+ and [Claude Code](https://docs.claude.com/en/docs/claude-code/overview) first
49
+ (see ["Requirements"](#requirements) below for exact versions). Nothing is
50
+ installed globally: `npx` and `pnpm dlx` fetch the CLI, run it once and leave
51
+ only a download cache behind.
52
+
53
+ **A new project.** Point the CLI at an empty or missing directory:
54
+
55
+ ```bash
56
+ # with npx
57
+ npx @monte3l/groundwork my-app
58
+
59
+ # or with pnpm
60
+ pnpm dlx @monte3l/groundwork my-app
61
+ ```
62
+
63
+ It writes the baseline, runs `git init` and `pnpm install`, and installs the
64
+ `/customize` skill. Add `--pack <name>` to install an optional
65
+ [pack](#packs) at the same time (`--list-packs` shows them).
66
+
67
+ **An existing project.** Run the CLI from the project's root, or pass its path.
68
+ It detects that the directory already looks like a project and switches to
69
+ adopt mode, which surveys it read-only and never overwrites a project file:
70
+
71
+ ```bash
72
+ cd path/to/your-project
73
+
74
+ # with npx
75
+ npx @monte3l/groundwork .
76
+
77
+ # or with pnpm
78
+ pnpm dlx @monte3l/groundwork .
79
+ ```
80
+
81
+ It writes a report and staged copies of what it would add under
82
+ `.groundwork/`, plus one guarded copy of the `/customize` skill.
83
+
84
+ **Then, in either case:**
85
+
86
+ ```bash
87
+ cd my-app # or stay in your existing project
88
+ claude # open the project in Claude Code
89
+ ```
90
+
91
+ Run `/customize` inside Claude Code. For a new project it interviews you and
92
+ tailors the baseline. For an existing one it first reconciles the survey
93
+ against your repo and confirms each change with you before editing anything.
55
94
 
56
95
  > [!NOTE]
57
96
  > Earlier releases shipped on other [dist-tags](https://github.com/monte3l/m3l-groundwork/blob/main/docs/glossary.md#dist-tag).
@@ -106,10 +145,13 @@ the wiring contract.
106
145
 
107
146
  | Pack | Contents |
108
147
  | ---------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
109
- | `harness-extras` | The compaction-handoff hook pair, a read-only Bash guard, and a five-row Claude Code status line plus a per-subagent row renderer, width-fit to the terminal. |
148
+ | `harness-extras` | The compaction-handoff hook pair and a read-only Bash guard. |
149
+ | `statusline` | A five-row Claude Code status line plus a per-subagent row renderer, width-fit to the terminal. Sets the `statusLine` and `subagentStatusLine` keys. |
110
150
  | `quality` | A per-file size ratchet gate over `src/` and `tests/`, and a read-only `type-design-analyzer` review agent. Adopt-capable, recommended for every project kind. |
111
- | `github` | Anthropic's official Claude Code GitHub Action (`@claude` mention-mode) plus an automated PR-review Action and three `gh`-CLI skills (Dependabot PR triage, scan-alert triage, PR-check watching) -- no hooks, no gate. Needs an auth secret the pack cannot create. |
112
- | `publishing` | A release pipeline (changesets version-PR / staged, provenance-attested npm publish via trusted publishing) plus SPDX license headers. Fresh mode only. |
151
+ | `claude-github` | Anthropic's official Claude Code GitHub Action as two workflows: `@claude` mention-mode and an automated PR-review comment on every pull request -- no hooks, no gate. Needs an auth secret the pack cannot create, and the review spends API credits per PR. |
152
+ | `github` | Three `gh`-CLI skills (Dependabot PR triage, scan-alert triage, PR-check watching) -- no hooks, no gate, no workflows. |
153
+ | `publishing` | A release pipeline (changesets version-PR / staged, provenance-attested npm publish via trusted publishing). Fresh mode only. |
154
+ | `licensing` | An SPDX/REUSE license-header gate (`check-license-headers.mjs`, a `lint`-group verify step) and a `REUSE.toml` template. Fresh mode only, because the one-time backfill fails an established project's `pnpm verify`. |
113
155
  | `supply-chain` | Secret scanning (`gitleaks.yml`) and an OpenSSF Scorecard run (`scorecard.yml`) for any GitHub project, published or not. Two workflows (read-only apart from the Scorecard job's `security-events` and `id-token` write), adopt-capable, recommended for every project kind. |
114
156
  | `worktrees` | Enforces that all src/tests development happens inside an isolated git worktree, on any branch, for any caller: a `working-in-worktrees` skill (start/status/sync/finish/fan-out), a dependency-install backstop hook, a guard stricter than the baseline's own branch/hub guards, and a `.worktreeinclude` copying `.env`/`.env.local`/`.env.*.local` into every worktree Claude Code creates. Changes the day-to-day workflow -- opt-in, not recommended by default. |
115
157
 
@@ -276,7 +318,7 @@ everything else that goes wrong at run time.
276
318
  | Unrecognized flag, missing flag value, repeated `--name`, missing or extra target argument | 2 | Prints the error and the usage text. |
277
319
  | `--adopt` with `--fresh` | 2 | The two are mutually exclusive. |
278
320
  | `--name` is not a valid npm package name, or a `--pack` value is not a valid pack name | 2 | Prints the rule it broke and the usage text. |
279
- | Unknown `--pack` | 2 | Lists the available packs. A retired name gets a hint: `statusline` was renamed to `harness-extras`. |
321
+ | Unknown `--pack` | 2 | Lists the available packs. |
280
322
  | `--pack` for a pack without fresh mode | 2 | Says the pack does not support fresh mode and lists its modes. |
281
323
  | Adopt mode with `--adopt` and a missing target directory, `--force` or `--pack` | 2 | Adopt never writes project files and surveys every pack itself; run `/customize` to install a pack. |
282
324
  | Windows, fresh mode | 1 | Refuses before writing anything: Linux and macOS only. Adopt mode, `--help`, `--version` and `--list-packs` still run. |
@@ -473,7 +473,7 @@ const descriptionSubstance = {
473
473
  else if (description.length > DESCRIPTION_MAX) {
474
474
  failures.push({
475
475
  subject,
476
- message: `description is ${description.length} chars; over ${DESCRIPTION_MAX} it is truncated in the skill listing`,
476
+ message: `description is ${description.length} chars; over the ${DESCRIPTION_MAX}-char Agent Skills maximum (Claude Code also truncates description + when_to_use at 1,536 in its skill listing)`,
477
477
  });
478
478
  }
479
479
  };
package/dist/main.js CHANGED
@@ -389,28 +389,18 @@ function describeInstallFailure(error) {
389
389
  return `\`pnpm install\` failed (${code})`;
390
390
  return "`pnpm install` failed";
391
391
  }
392
- // A pack that was renamed or folded into another, mapped to its successor so
393
- // an old `--pack` name gets a pointed hint rather than a bare "unknown pack".
394
- const RENAMED_PACKS = new Map([
395
- ["statusline", "harness-extras"],
396
- ]);
397
392
  /**
398
393
  * Loads one `--pack` for fresh mode -- called for every pack before any file
399
- * is written. An unknown name (with a rename hint when {@link RENAMED_PACKS}
400
- * knows its successor) or a fresh-incompatible pack is a usage error (exit
401
- * 2); a pack that exists but whose manifest fails to load propagates
402
- * `loadPack`'s own error unchanged (exit 1), since that is a broken install,
403
- * not a bad invocation.
394
+ * is written. An unknown name (the message lists the available packs) or a
395
+ * fresh-incompatible pack is a usage error (exit 2); a pack that exists but
396
+ * whose manifest fails to load propagates `loadPack`'s own error unchanged
397
+ * (exit 1), since that is a broken install, not a bad invocation.
404
398
  */
405
399
  function resolveFreshPack(name) {
406
400
  const available = listPackNames();
407
401
  if (!available.includes(name)) {
408
- const successor = RENAMED_PACKS.get(name);
409
- const hint = successor === undefined
410
- ? ""
411
- : ` -- it was renamed to "${successor}"; use --pack ${successor}`;
412
402
  const list = available.length > 0 ? available.join(", ") : "none";
413
- throw new CliUsageError(`unknown pack "${name}"${hint} (available: ${list})\n\n${USAGE}`);
403
+ throw new CliUsageError(`unknown pack "${name}" (available: ${list})\n\n${USAGE}`);
414
404
  }
415
405
  const pack = loadPack(name);
416
406
  if (!pack.manifest.modes.includes("fresh")) {
@@ -46,6 +46,8 @@ interface SettingsHookCommand {
46
46
  command: string;
47
47
  if?: string;
48
48
  timeout?: number;
49
+ /** Hook failure policy (Claude Code 2.1.295+); defaults to `"continue"`. */
50
+ onFailure?: "continue" | "block";
49
51
  }
50
52
  interface SettingsHookEntry {
51
53
  matcher?: string;
@@ -58,7 +60,7 @@ export type SettingsHooksFragment = Record<string, SettingsHookEntry[]>;
58
60
  * counts as a match -- e.g. `PreCompact`, which has none); within a matched
59
61
  * entry, only `hooks` commands not already present (compared by exact
60
62
  * `command` string) are appended. A command that matches on `command` but
61
- * differs in its other fields (`if`, `timeout`) is a hard collision, never
63
+ * differs in its other fields (`if`, `timeout`, `onFailure`) is a hard collision, never
62
64
  * a silent overwrite. An event name of `__proto__`, `constructor` or
63
65
  * `prototype` throws: the check runs per key inside the loop, before that
64
66
  * key is read or written, and nothing escapes the failed merge -- `existing`
@@ -110,7 +110,7 @@ function assertSafeKey(key, context) {
110
110
  * counts as a match -- e.g. `PreCompact`, which has none); within a matched
111
111
  * entry, only `hooks` commands not already present (compared by exact
112
112
  * `command` string) are appended. A command that matches on `command` but
113
- * differs in its other fields (`if`, `timeout`) is a hard collision, never
113
+ * differs in its other fields (`if`, `timeout`, `onFailure`) is a hard collision, never
114
114
  * a silent overwrite. An event name of `__proto__`, `constructor` or
115
115
  * `prototype` throws: the check runs per key inside the loop, before that
116
116
  * key is read or written, and nothing escapes the failed merge -- `existing`
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@monte3l/groundwork",
3
- "version": "1.0.4",
3
+ "version": "1.0.5",
4
4
  "description": "CLI that bootstraps a TypeScript project with a strict toolchain and Claude Code setup, or read-only surveys an existing one, plus a /customize plugin that tailors it to current guidance.",
5
5
  "keywords": [
6
6
  "bootstrap",
@@ -61,18 +61,21 @@ confirmed in Step 0.4:
61
61
  `recommendPacks(answers)` from `pack-map.ts` (alongside this file, same
62
62
  copy mechanism as `kind-facet-map.ts`) with the now-confirmed
63
63
  `InterviewAnswers` and compare its verdict against Step 0.4's decision.
64
- `harness-extras`, `github`, `supply-chain`, `quality`, and `worktrees`
65
- never disagree (none of the five's recommendation varies by kind or
66
- answer -- `worktrees` is always `recommended: false`, since it changes the
67
- day-to-day workflow rather than adding a nicety, and the other four are
68
- always `recommended: true`), but `publishing`'s does (recommended for
69
- `library`/`cli`, not for `frontend`/`service`) — if the comparison surfaces a real
64
+ `harness-extras`, `statusline`, `github`, `claude-github`, `supply-chain`,
65
+ `quality`, and `worktrees` never disagree (none of the seven's
66
+ recommendation varies by kind or answer -- `worktrees` is always
67
+ `recommended: false`, since it changes the day-to-day workflow rather than
68
+ adding a nicety, and `claude-github` is always `recommended: false`, since
69
+ its workflows need an auth secret and the review spends API credits on
70
+ every PR; the other five are always `recommended: true`), but `publishing`'s and `licensing`'s do (each
71
+ recommended for `library`/`cli`, not for `frontend`/`service`) — if the comparison surfaces a real
70
72
  conflict there or for any future kind-scoped pack, raise it rather than
71
73
  silently overriding the user's Step 0.4 answer, mirroring Step 4's "the
72
74
  one exception" rule for guidance findings. Before installing, check the
73
75
  pack's own `modes`: a pack whose `modes` doesn't include `"adopt"` (today,
74
76
  `publishing` — its release flow encodes decisions too project-specific to
75
- apply blind) is never auto-installed here even if staged and approved;
77
+ apply blind — and `licensing`, whose one-time SPDX header backfill fails an
78
+ established project's `pnpm verify`) is never auto-installed here even if staged and approved;
76
79
  instead, state in Step 7 that it needs a manual install (point at `.groundwork/packs/<name>/` and the pack's own `adoptNotes`, and say to strip the `.staged` suffix from every name when copying by hand and to replace every `__KEY__` token with the project's real value) and stop
77
80
  there for that pack. Before merging any staged `pack.json.staged` wiring, repeat
78
81
  the prototype-key check from Step 0.4(c): if a key of `wiring.settings`,
@@ -90,20 +93,23 @@ confirmed in Step 0.4:
90
93
  from `pack.json.staged` (the bytes Step 0.1 read) for a schema 5 or later inventory and
91
94
  from the unsuffixed `pack.json` for schema 1-4,
92
95
  against what Step 0.2's deep read already found — a `.claude/settings.json`
93
- hook fragment merges the same way the baseline's own hook entries would;
96
+ hook fragment merges the same way the baseline's own hook entries would, and when the project already registers the same hook `command` (a pack installed before it gained a key such as `onFailure`), add only the keys the fragment has and the existing registration lacks, keep every field the project already sets, and ask before changing a key both define with a different value (the CLI's `mergeSettingsHooks` never overwrites one);
94
97
  `wiring.settingsTopLevel` is a set of top-level keys (e.g. `statusLine`)
95
98
  planted whole, and only when the project doesn't already define that key —
96
99
  when it does, show the existing value and ask, since the CLI's own
97
100
  `mergeSettingsTopLevel` treats a differing value as a hard error and this
98
101
  hand-applied path must not be laxer than the automated one (also check
99
102
  `.claude/settings.local.json` and the user's `~/.claude/settings.json`,
100
- either of which can shadow a project `statusLine`). For `harness-extras`
103
+ either of which can shadow a project `statusLine`). For `statusline`
101
104
  specifically, a `statusLine`/`subagentStatusLine` collision is never a
102
- reason to fail the whole pack install: skip just those two settings keys
103
- and the three statusline scripts (`statusline.mjs`, `statusline-layout.mjs`,
104
- `subagent-statusline.mjs`) and install the pack's other artifacts (the
105
- compaction-handoff hooks and `guard-readonly-bash`) normally, stating the
106
- skip plainly in Step 7. `guard-readonly-bash.mjs` imports
105
+ reason to fail the whole pack install: skip just the colliding key and the
106
+ script that renders it (`statusline.mjs` for `statusLine`,
107
+ `subagent-statusline.mjs` for `subagentStatusLine`; `statusline-layout.mjs`
108
+ stays while either remains), install the other key normally, and skip the
109
+ whole pack only when both keys are already set, stating the skip plainly in
110
+ Step 7. The pack is separate from `harness-extras` for exactly this reason,
111
+ so a skipped status line never costs the compaction-handoff hooks or
112
+ `guard-readonly-bash`. For `harness-extras`, `guard-readonly-bash.mjs` imports
107
113
  `bin/lib/agent-roster.mjs`: when the project keeps no agents and that file is
108
114
  not being installed, skip that one hook too (the pack's `adoptNotes` allow
109
115
  it), record it in `round1.dropped` and say in Step 7 that it was skipped and
@@ -142,9 +148,11 @@ any reason goes into `round1.dropped` with that reason.
142
148
  `PluginRecommendationContext`: `chosenPacks` is whatever the packs decision
143
149
  just above actually landed on (fresh: what `--pack` installed at bootstrap
144
150
  time, detectable from the installed files -- e.g. `.changeset/config.json`
145
- means `publishing`, `.github/workflows/claude-pr-review.yml` means `github`
146
- (the only marker `plugin-map.ts` actually reads; `harness-extras` has no
147
- plugin that varies by its presence, so it needs no marker here); adopt: the
151
+ means `publishing`, `.claude/skills/watching-pr-checks/SKILL.md` means `github`
152
+ (the only marker `plugin-map.ts` actually reads; the `claude-github` pack's
153
+ `.github/workflows/claude-pr-review.yml` needs no marker, since no plugin
154
+ varies by its presence; `harness-extras` and `statusline` have
155
+ no plugin that varies by their presence, so they need no marker here); adopt: the
148
156
  packs Step 0.4(c) confirmed, revised by this step's own `recommendPacks`
149
157
  comparison if it changed anything); `hasCustomSkills` is `false` for a fresh
150
158
  bootstrap (nothing has authored a skill yet) and, for adopt mode, whatever
@@ -88,7 +88,7 @@ export const TYPESCRIPT_DOMAIN_GLOBS = [
88
88
  * `bin/lib/*.mjs` are typescript-domain globs, so without this list the
89
89
  * grader's own rules would be swept by `typescript-guidance` -- wrong,
90
90
  * because they encode Claude Code harness guidance. The Claude Code Action
91
- * workflows (`claude.yml`, and the `github` pack's automated-review
91
+ * workflows (`claude.yml`, and the `claude-github` pack's automated-review
92
92
  * `claude-pr-review.yml`) are here for the same reason against
93
93
  * `.github/workflows/*.yml`: their trigger, action pin and model choice are
94
94
  * Anthropic guidance, not toolchain. Consulted before the typescript list in
@@ -19,43 +19,61 @@ export interface PackRecommendation {
19
19
  }
20
20
 
21
21
  /**
22
- * `harness-extras`'s remaining artifacts (the compaction-handoff hook pair
23
- * and a read-only Bash guard) are harness-level, not domain-level -- they
22
+ * `harness-extras`'s remaining artifacts (the compaction-handoff hook pair,
23
+ * which carries a work summary across a compaction, and a read-only Bash
24
+ * guard for review spokes) are harness-level, not domain-level -- they
24
25
  * apply to any TypeScript project regardless of what it's building. The
25
26
  * type-design-analyzer agent and the file-budget gate it once also carried
26
- * now live in the separate `quality` pack. Its folded-in
27
- * statusLine scripts read only the stdin payload, `.git/HEAD` (via
28
- * `node:fs`, never a `git` subprocess) and `os.freemem()`/`os.totalmem()`,
29
- * and `statusLine` is the only documented surface carrying live
30
- * `context_window.used_percentage`: no hook event receives token or context
31
- * data, so "when to compact" can live nowhere else.
27
+ * now live in the separate `quality` pack, and its status-line scripts in
28
+ * the separate `statusline` pack.
32
29
  */
33
30
  function recommendHarnessExtras(): PackRecommendation {
34
31
  return {
35
32
  name: "harness-extras",
36
33
  recommended: true,
37
34
  because:
38
- "its compaction-handoff hooks (which carry work across a context " +
39
- "compaction) and read-only Bash guard are harness-level, not tied " +
40
- "to any particular project kind; its " +
41
- "statusLine is the only surface that exposes live context-window " +
42
- "pressure -- no hook event receives token data -- and reads only the " +
43
- "stdin payload plus local git and memory state. It does occupy five " +
44
- "terminal rows (session, model, context, quota, work).",
35
+ "its compaction-handoff hooks (which carry a work summary across a " +
36
+ "compaction) and its read-only Bash guard for review spokes are " +
37
+ "harness-level, not tied to any particular project kind.",
45
38
  };
46
39
  }
47
40
 
48
41
  /**
49
- * `github`'s workflow assumes only GitHub, which this baseline's whole
50
- * toolchain (CI, Dependency Review, rulesets) already does -- nothing about
51
- * any project kind. Beyond the mention-mode Action (`claude.yml`) it ships a
52
- * second workflow, `claude-pr-review.yml`, which posts an automated Claude
53
- * review comment on every PR, plus three GitHub-operations skills:
42
+ * `statusline` is the five-row `statusLine` script (session, model,
43
+ * context, quota, work) plus a per-subagent row renderer, carved out of
44
+ * `harness-extras`. Its scripts read only the stdin payload, `.git/HEAD`
45
+ * (via `node:fs`, never a `git` subprocess) and
46
+ * `os.freemem()`/`os.totalmem()`, and `statusLine` is the only documented
47
+ * surface carrying live `context_window.used_percentage`: no hook event
48
+ * receives token or context data, so "when to compact" can live nowhere
49
+ * else. That is harness-level, so it is recommended for every project
50
+ * kind. In adopt mode a project that already defines `statusLine` or
51
+ * `subagentStatusLine` keeps its own value -- it is never overwritten, so
52
+ * the owner decides which one to run.
53
+ */
54
+ function recommendStatusline(): PackRecommendation {
55
+ return {
56
+ name: "statusline",
57
+ recommended: true,
58
+ because:
59
+ "its five-row statusLine (session, model, context, quota, work) plus " +
60
+ "a per-subagent row renderer is the only surface that exposes live " +
61
+ "context-window pressure -- no hook event receives token data -- and " +
62
+ "reads only the stdin payload plus local git and memory state. It " +
63
+ "does occupy five terminal rows. A project that already defines " +
64
+ "statusLine or subagentStatusLine keeps its own value (it is never " +
65
+ "overwritten) and must decide which one to run.",
66
+ };
67
+ }
68
+
69
+ /**
70
+ * `github` is three GitHub-operations skills driven by the `gh` CLI:
54
71
  * `reviewing-dependabot-prs` (classifies and batch-merges open Dependabot
55
72
  * PRs), `triaging-scan-alerts` (triages open code-scanning alerts to
56
- * file:line) and `watching-pr-checks`. It is not zero-setup: it adds two
57
- * GitHub Actions workflows and needs an auth secret the pack cannot create
58
- * itself.
73
+ * file:line) and `watching-pr-checks`. It assumes only GitHub, which this
74
+ * baseline's whole toolchain (CI, Dependency Review, rulesets) already does
75
+ * -- nothing about any project kind -- and needs only `gh` CLI auth, so it
76
+ * is recommended for every kind.
59
77
  */
60
78
  function recommendGithub(): PackRecommendation {
61
79
  return {
@@ -63,22 +81,43 @@ function recommendGithub(): PackRecommendation {
63
81
  recommended: true,
64
82
  because:
65
83
  "the baseline's toolchain already assumes GitHub (CI, Dependency " +
66
- "Review, rulesets), so a Claude Code Action integration applies to " +
67
- "any project kind. Alongside the mention-mode Action it adds a " +
68
- "second workflow (claude-pr-review.yml) that posts an automated " +
69
- "Claude review comment on every PR, plus skills for classifying and " +
70
- "batch-merging open Dependabot PRs (reviewing-dependabot-prs), " +
71
- "triaging open code-scanning alerts to file:line " +
72
- "(triaging-scan-alerts) and watching a PR's checks " +
73
- "(watching-pr-checks). It does add two GitHub Actions workflows and " +
74
- "needs an auth secret configured by hand -- the pack cannot create it.",
84
+ "Review, rulesets), so its gh-CLI skills apply to any project kind: " +
85
+ "classifying and batch-merging open Dependabot PRs " +
86
+ "(reviewing-dependabot-prs), triaging open code-scanning alerts to " +
87
+ "file:line (triaging-scan-alerts) and watching a PR's checks " +
88
+ "(watching-pr-checks). It needs only gh CLI auth.",
75
89
  };
76
90
  }
77
91
 
78
92
  /**
79
- * `publishing` is the one kind-scoped pack: a release pipeline (a changesets
93
+ * `claude-github` is Anthropic's Claude Code GitHub Action as two
94
+ * workflows: `claude.yml` (mention-mode, answering `@claude` in issues and
95
+ * PRs) and `claude-pr-review.yml` (an automated Claude review comment on
96
+ * every PR). It needs an auth secret (`ANTHROPIC_API_KEY` or a Claude
97
+ * subscription OAuth token) configured by hand, since the pack cannot
98
+ * create it, and the review workflow spends API credits on every PR. That
99
+ * is a cost and setup decision no interview answer can infer, so it is
100
+ * opt-in for every project kind -- but it stays on offer.
101
+ */
102
+ function recommendClaudeGithub(): PackRecommendation {
103
+ return {
104
+ name: "claude-github",
105
+ recommended: false,
106
+ because:
107
+ "it adds Anthropic's Claude Code GitHub Action as two workflows: " +
108
+ "@claude mention-mode (claude.yml) and an automated Claude review " +
109
+ "comment on every PR (claude-pr-review.yml). It needs an auth secret " +
110
+ "(ANTHROPIC_API_KEY or a Claude subscription OAuth token) configured " +
111
+ "by hand -- the pack cannot create it -- and the review workflow " +
112
+ "spends API credits on every PR, so it is opt-in.",
113
+ };
114
+ }
115
+
116
+ /**
117
+ * `publishing` is a kind-scoped pack: a release pipeline (a changesets
80
118
  * version PR, then a staged, provenance-attested npm publish via trusted
81
- * publishing) plus SPDX/REUSE license headers. Secret scanning and OpenSSF
119
+ * publishing) plus a gate that the published type declarations' dependencies
120
+ * are installable by consumers (dts-deps). Secret scanning and OpenSSF
82
121
  * Scorecard are not part of it -- they are the separate `supply-chain` pack,
83
122
  * since they apply to any repo, published or not. A `library` or `cli`
84
123
  * typically publishes an npm package; a
@@ -95,12 +134,13 @@ function recommendPublishing(answers: InterviewAnswers): PackRecommendation {
95
134
  because: recommended
96
135
  ? "a library or CLI typically ships an npm package, so this pack's " +
97
136
  "release pipeline applies: a changesets version PR, then a staged, " +
98
- "provenance-attested npm publish via trusted publishing -- plus " +
99
- "SPDX/REUSE license headers. The repository-hygiene scans that " +
100
- "apply to any project, published or not, are the separate " +
101
- "supply-chain pack. It is fresh-mode only and needs one-time npm and GitHub setup (the " +
102
- "trusted publisher, the release credentials) the pack cannot do " +
103
- "itself."
137
+ "provenance-attested npm publish via trusted publishing, with a " +
138
+ "dts-deps gate that the published type declarations' dependencies " +
139
+ "resolve for consumers. The repository-hygiene scans that apply to " +
140
+ "any project, published or not, are the separate supply-chain " +
141
+ "pack. It is fresh-mode only and needs one-time npm and GitHub " +
142
+ "setup (the trusted publisher, the release credentials) the pack " +
143
+ "cannot do itself."
104
144
  : "a frontend or service is typically deployed, and a Claude Code " +
105
145
  "plugin ships through a marketplace manifest, rather than published " +
106
146
  "to a package registry, so a changesets/npm release pipeline is not " +
@@ -109,6 +149,38 @@ function recommendPublishing(answers: InterviewAnswers): PackRecommendation {
109
149
  };
110
150
  }
111
151
 
152
+ /**
153
+ * `licensing` is the half carved out of `publishing`: an SPDX/REUSE
154
+ * license-header gate (`check-license-headers`) plus a `REUSE.toml`
155
+ * template. Licensing metadata matters most for code that is distributed,
156
+ * so it is pre-selected for a `library` or `cli`, the same kinds
157
+ * `publishing` targets -- but it is independent of `publishing` and stays
158
+ * on offer for every kind. It is fresh-mode only: an established project
159
+ * fails `pnpm verify` until the one-time backfill (`git add -A`, then
160
+ * `node bin/check-license-headers.mjs --fix`) has touched every file.
161
+ */
162
+ function recommendLicensing(answers: InterviewAnswers): PackRecommendation {
163
+ const recommended = answers.kind === "library" || answers.kind === "cli";
164
+ return {
165
+ name: "licensing",
166
+ recommended,
167
+ because: recommended
168
+ ? "a library or CLI is distributed, which is where licensing metadata " +
169
+ "matters most, so this pack's SPDX/REUSE license-header gate and " +
170
+ "REUSE.toml template apply. It is independent of the publishing " +
171
+ "pack. It is fresh-mode only: the one-time backfill (git add -A, " +
172
+ "then node bin/check-license-headers.mjs --fix) has to touch every " +
173
+ "file before pnpm verify passes, which an established project " +
174
+ "would fail until then."
175
+ : "a frontend, service or Claude Code plugin is less often " +
176
+ "redistributed as source, so an SPDX/REUSE license-header gate is " +
177
+ "not recommended by default -- it is still available, independent " +
178
+ "of the publishing pack. It is fresh-mode only: the one-time " +
179
+ "backfill (git add -A, then node bin/check-license-headers.mjs " +
180
+ "--fix) has to touch every file before pnpm verify passes.",
181
+ };
182
+ }
183
+
112
184
  /**
113
185
  * `worktrees` is opt-in for every project kind: rather than adding a
114
186
  * nicety on top of the existing workflow, it changes the workflow itself --
@@ -170,18 +242,27 @@ function recommendQuality(): PackRecommendation {
170
242
  }
171
243
 
172
244
  /**
173
- * Every pack's recommendation for the given interview answers. Only
174
- * `publishing` varies by `answers.kind` today; `harness-extras`, `github`,
175
- * `supply-chain` and `quality` are recommended for every kind, and
176
- * `worktrees` is opt-in for every kind.
245
+ * Every pack's recommendation for the given interview answers. Two packs
246
+ * vary by `answers.kind` today, `publishing` and `licensing`;
247
+ * `harness-extras`, `statusline`, `github`, `supply-chain` and `quality`
248
+ * are recommended for every kind, and `claude-github` and `worktrees` are
249
+ * opt-in for every kind.
250
+ *
251
+ * @example
252
+ * ```ts
253
+ * const picks = recommendPacks(answers).filter((r) => r.recommended);
254
+ * ```
177
255
  */
178
256
  export function recommendPacks(
179
257
  answers: InterviewAnswers,
180
258
  ): PackRecommendation[] {
181
259
  return [
182
260
  recommendHarnessExtras(),
261
+ recommendStatusline(),
183
262
  recommendGithub(),
263
+ recommendClaudeGithub(),
184
264
  recommendPublishing(answers),
265
+ recommendLicensing(answers),
185
266
  recommendWorktrees(),
186
267
  recommendSupplyChain(),
187
268
  recommendQuality(),
@@ -20,7 +20,7 @@
20
20
  * again:
21
21
  *
22
22
  * - `code-review` -- duplicates the built-in `/code-review` and the optional
23
- * `github` pack's PR-review Action.
23
+ * `claude-github` pack's PR-review Action.
24
24
  * - `code-simplifier` -- duplicates the built-in `/simplify`; its
25
25
  * unrestricted-tools agent is blocked by the baseline's hub-and-spoke
26
26
  * write guard on `src/`/`tests/` anyway.
@@ -11,6 +11,10 @@
11
11
  * instead of being discovered when `guard-branch-isolation.mjs` blocks a
12
12
  * src/test write on `main`.
13
13
  *
14
+ * It sees the line as typed: Claude Code runs `UserPromptSubmit` hooks before
15
+ * a skill or slash command expands (2.1.299), so `/creating-prs fix the lint`
16
+ * is judged on those words, and a bare `/starting-work` carries no change verb.
17
+ *
14
18
  * It is advisory: it emits `additionalContext` and exits 0. It never blocks.
15
19
  * The heuristic is deliberately loose (inject on any change-intent verb,
16
20
  * stay quiet on obvious reads) -- a spurious reminder is cheap; a missed one
@@ -36,6 +36,52 @@ auto, `acceptEdits` or `bypassPermissions` mode overrides a spoke's
36
36
  guards bound a spoke, not `permissionMode`. Source:
37
37
  [sub-agents](https://code.claude.com/docs/en/sub-agents).
38
38
 
39
+ ## Guard limits and overrides (detail)
40
+
41
+ Sources: [hooks](https://code.claude.com/docs/en/hooks),
42
+ [hooks guide](https://code.claude.com/docs/en/hooks-guide),
43
+ [managed settings](https://code.claude.com/docs/en/managed-settings),
44
+ [mods events](https://code.claude.com/docs/en/plugins/mods/events),
45
+ [sub-agents](https://code.claude.com/docs/en/sub-agents) and
46
+ [agent teams](https://code.claude.com/docs/en/agent-teams).
47
+
48
+ - **The `Bash` screen is conservative.** `guard-hub-src-writes.mjs` blocks
49
+ redirects, `tee`, `sed -i`, `cp`/`mv` into a guarded path, an interpreter
50
+ snippet that writes there, and a copy, move or removal of a parent
51
+ (`rsync -a /tmp/x/ ./`, `rm -rf .`). It cannot see an interpreter running a
52
+ script from a project file, an `eval`, a computed path, a parent of the
53
+ project root (`rm -rf ..`), a workspace container below the project root, a
54
+ build step, a formatter or fixer (`prettier --write`, `eslint --fix`),
55
+ `git rm`/`git mv`, `find -exec`, `xargs` with a computed operand,
56
+ `tar`/`curl -o`, or a very long interpreter call (allowed with a stderr
57
+ note).
58
+ - **Managed settings outrank every project file.** A `managed-settings.json`
59
+ file, an MDM policy or a console-managed policy has no project-level
60
+ override, and `allowManagedHooksOnly` and `allowManagedPermissionRulesOnly`
61
+ mean only managed hooks run and only managed permission rules apply. The
62
+ file lives outside any working tree, so nothing here can detect it. On a
63
+ machine you don't control, run `/status` and read its "Setting sources"
64
+ line: if `guard-hub-src-writes.mjs` and `guard-branch-isolation.mjs` aren't
65
+ running, treat hub-and-spoke as an unenforced checklist.
66
+ - **An installed mod sits below managed settings and above project hooks.** A
67
+ mod that handles `tool.check` can approve a call a non-managed `PreToolUse`
68
+ hook blocked, and project `PreToolUse` hooks run only after the last mod
69
+ calls `next`. Only managed hooks outrank a mod. Check `/plugin` alongside
70
+ `/status`.
71
+ - **Who counts as a writer.** The guard needs a writer-spoke `agent_type` and a
72
+ non-empty `agent_id`. A fork subagent (on by default in every surface since
73
+ Claude Code 2.1.302) is not a writer spoke, so unless its payload reports a
74
+ writer `agent_type` it is treated as the hub (the docs don't say what a
75
+ fork's payload carries). A fork cannot spawn further agents. With agent teams on, an in-process
76
+ teammate has a documented `agent_id`; one built from a writer definition that
77
+ also reports that `agent_type` passes as a writer. A split-pane teammate does
78
+ not get `disallowedTools` (only an in-process one does), so every spoke's
79
+ `tools` list also omits `Agent`.
80
+ - **What `onFailure: "block"` covers.** A failure is a hook that cannot start,
81
+ times out, exits with any code other than 0 or 2, or prints invalid JSON. It
82
+ has no effect on `Stop`, `SubagentStop`, `TaskCompleted`, `TeammateIdle` or
83
+ async hooks.
84
+
39
85
  ## Dispatch sizing
40
86
 
41
87
  - **Decompose before you dispatch.** Scale the dispatch to task complexity —
@@ -87,73 +87,54 @@ Skills, each in `.claude/skills/<name>/`:
87
87
  sweeps (see Freshness).
88
88
 
89
89
  **The guard also screens `Bash`, but only conservatively.** The same hook
90
- runs on every `Bash` call and blocks a command that visibly writes into
91
- `src/` or `tests/` (redirects, `tee`, `sed -i`, `cp`/`mv` into them, a
92
- `python`/`node`/`php` snippet that writes or deletes there, a copy, move or
93
- removal of a parent of `src/` such as `rsync -a /tmp/x/ ./` or `rm -rf .`, and
94
- similar) unless the caller is `test-author` or `code-implementer`. It cannot
95
- catch an indirect write -- an interpreter running a script from a project
96
- file, an `eval`, a computed path, a parent of the project root (`rm -rf ..`), a workspace container below the project root (`tools/packages/foo`), a build step, a formatter or fixer
97
- (`prettier --write`, `eslint --fix`), `git rm`/`git mv`, `find -exec`, `xargs` with a computed operand, `tar`/`curl -o`, a very long
98
- interpreter call (allowed with a stderr note) -- so hub-and-spoke remains a convention backed by a guard that raises the bar,
99
- not a proof. To override it deliberately, run the command yourself with the
100
- `!` prefix at the Claude Code prompt, or edit the hook's registration in
101
- `.claude/settings.json`.
102
-
103
- **A Claude Code Enterprise/managed deployment sits above this and can
104
- silently disable it.** Managed settings (a `managed-settings.json` file, an
105
- MDM policy, or a claude.ai-console-managed remote policy) take precedence
106
- over every file this baseline installs, with no project-level override, and
107
- two managed-only keys -- `allowManagedHooksOnly` and
108
- `allowManagedPermissionRulesOnly` -- mean that only managed hooks run
109
- (`allowManagedHooksOnly`) and only managed permission rules apply
110
- (`allowManagedPermissionRulesOnly`) (see
111
- [Claude Code's managed-settings docs](https://code.claude.com/docs/en/managed-settings)).
112
- Nothing in this project can detect or gate against that from inside the
113
- repo -- the managed file lives outside any working tree, at an OS-level
114
- path. Run `/status` on a machine you don't control before trusting
115
- `guard-hub-src-writes.mjs`/`guard-branch-isolation.mjs`: its "Setting
116
- sources" line names every active source, and if these hooks aren't among
117
- what's actually running, treat hub-and-spoke as an unenforced checklist
118
- until confirmed otherwise.
119
-
120
- **An installed Claude Code mod is a second silent override, below managed
121
- settings.** Per the
122
- [hooks guide](https://code.claude.com/docs/en/hooks-guide), a mod that
123
- handles `tool.check` "can approve a call that your `PreToolUse` hook blocked,
124
- unless the hook is in managed settings", and project `PreToolUse` hooks run
125
- only after the last mod calls `next` (see the
126
- [mods events reference](https://code.claude.com/docs/en/plugins/mods/events)).
127
- Only managed-settings hooks outrank a mod, and nothing in the repo can detect
128
- one. Check `/plugin` for installed mods alongside `/status`.
90
+ blocks a `Bash` command that visibly writes into `src/` or `tests/` (redirects,
91
+ `tee`, `sed -i`, `cp`/`mv` into them, an interpreter snippet that writes there,
92
+ a copy, move or removal of a parent of `src/` such as `rm -rf .`) unless the caller is
93
+ `test-author` or `code-implementer`. It cannot catch an indirect write (a
94
+ script from a project file, an `eval`, a computed path, a build step, a
95
+ formatter or fixer, `git rm`/`git mv`, `tar`/`curl -o`), so hub-and-spoke is a
96
+ convention backed by a guard that raises the bar, not a proof. To override it
97
+ deliberately, run the command yourself with the `!` prefix at the Claude Code
98
+ prompt, or edit the hook's registration in `.claude/settings.json`.
99
+
100
+ **Managed settings and mods can silently disable the guards.** A managed
101
+ deployment (`managed-settings.json`, MDM, a console policy) outranks every file
102
+ here, and `allowManagedHooksOnly` runs only managed hooks. An installed mod
103
+ that handles `tool.check` can approve a call a project `PreToolUse` hook
104
+ blocked, and only managed hooks outrank a mod. Nothing in the repo can detect
105
+ either: on a machine you don't control, check `/status` ("Setting sources") and
106
+ `/plugin` before trusting `guard-hub-src-writes.mjs` and
107
+ `guard-branch-isolation.mjs`. Detail and sources:
108
+ `.claude/rules/agent-dispatch.md`.
129
109
 
130
110
  **The guard keys on `agent_id` as well as `agent_type`.** A call counts as a
131
- writer spoke only with both present, so a call carrying `agent_type` but no
132
- `agent_id` is treated as the hub and is blocked.
133
-
134
- **Four guards block when they crash, and a secrets deny rule ships in
111
+ writer spoke only with both present, so one carrying `agent_type` alone is
112
+ treated as the hub and blocked, as is a fork subagent unless its payload
113
+ reports a writer `agent_type` with an `agent_id` (details and versions in
114
+ `.claude/rules/agent-dispatch.md`). An in-process
115
+ agent-teams teammate has a documented `agent_id`, so one built from a writer
116
+ definition that reports that `agent_type` would pass.
117
+
118
+ **Four guards block when they fail, and a secrets deny rule ships in
135
119
  `settings.json`.** `guard-secret-writes`, `guard-hub-src-writes`,
136
120
  `guard-branch-isolation` and `guard-protected-paths` carry
137
- `"onFailure": "block"` (Claude Code 2.1.295 or later). By default a hook that
138
- crashes or times out lets the call through; with this key it is blocked. It
139
- applies only to a failure the hook does not handle itself: each guard still
140
- catches bad input and exits 0 on purpose, so the key covers a load error, a
141
- missing `node` or the 30-second timeout. Because `guard-hub-src-writes` is also
142
- registered on `Bash`, a missing `node` or a timed-out scan blocks every `Bash`
143
- call until it is fixed (run the command yourself with the `!` prefix to get
144
- past it). The style guards and `guard-git-push-signed` stay fail-open on
145
- purpose: the push guard has its own command parser and the remote's
146
- signed-commit rule is the authoritative layer. `permissions.deny` blocks
147
- `Read(./.env)`, `Read(./.env.*)` and `Read(./secrets/**)`, which also covers
148
- `cat`, `head` and `tail` on those files in `Bash`. It does not cover `grep -r`
149
- or a script that opens the file itself, and nothing else in this baseline does.
121
+ `"onFailure": "block"` (Claude Code 2.1.295 or later): a hook that cannot
122
+ start, times out, exits with any code other than 0 or 2, or prints invalid
123
+ JSON blocks the call instead of letting it through. Each guard still handles
124
+ bad input itself and exits 0 on purpose. Because `guard-hub-src-writes` is also
125
+ on `Bash`, a missing `node` blocks every `Bash` call until fixed (run the
126
+ command with the `!` prefix to get past it). The style guards and
127
+ `guard-git-push-signed` stay fail-open on purpose: the push guard has its own
128
+ parser and the remote's signed-commit rule is the authoritative layer.
129
+ `permissions.deny` blocks `Read(./.env)`, `Read(./.env.*)` and
130
+ `Read(./secrets/**)`, which also covers `cat`, `head` and `tail` on them in
131
+ `Bash`. It does not cover `grep -r` or a script that opens the file itself.
150
132
 
151
133
  **A spoke's `permissionMode` can be overridden.** A session in auto,
152
134
  `acceptEdits` or `bypassPermissions` mode overrides a spoke's `permissionMode`
153
- (the field applies only when the session is in `default`, `dontAsk` or
154
- `plan`), so treat the hooks, not `permissionMode`, as what bounds a spoke.
155
- Claude Code also lets a subagent spawn subagents three layers deep by default;
156
- `disallowedTools: Agent` on every spoke is what keeps this roster flat.
135
+ (the field applies only in `default`, `dontAsk` or `plan`), so the hooks, not
136
+ `permissionMode`, bound a spoke. A subagent can spawn subagents three layers
137
+ deep by default; `disallowedTools: Agent` on every spoke keeps the roster flat.
157
138
 
158
139
  Full dispatch-sizing and recovery guidance: `.claude/rules/agent-dispatch.md`
159
140
  (auto-loads when editing `.claude/skills/**` or `.claude/agents/**`).
@@ -12,7 +12,9 @@
12
12
  * types. Skips cleanly (exit 0, one warning) when the target has no
13
13
  * `exports` field -- a project that isn't published doesn't need this gate,
14
14
  * and `/customize` removes the step entirely for non-library project kinds
15
- * rather than leaving a permanently-skipped one behind.
15
+ * rather than leaving a permanently-skipped one behind. attw resolves types
16
+ * with the TypeScript it bundles, not the project's own, so its verdict
17
+ * follows that resolver.
16
18
  */
17
19
  import process from "node:process";
18
20
  import { execFileSync } from "node:child_process";
@@ -691,7 +691,7 @@ const descriptionSubstance = {
691
691
  } else if (description.length > DESCRIPTION_MAX) {
692
692
  failures.push({
693
693
  subject,
694
- message: `description is ${description.length} chars; over ${DESCRIPTION_MAX} it is truncated in the skill listing`,
694
+ message: `description is ${description.length} chars; over the ${DESCRIPTION_MAX}-char Agent Skills maximum (Claude Code also truncates description + when_to_use at 1,536 in its skill listing)`,
695
695
  });
696
696
  }
697
697
  };
@@ -65,6 +65,8 @@ export default defineConfig(
65
65
  "@typescript-eslint/no-non-null-assertion": "error",
66
66
  "@typescript-eslint/no-floating-promises": "error",
67
67
  "@typescript-eslint/switch-exhaustiveness-check": "error",
68
+ // Overlaps `verbatimModuleSyntax` (typescript-eslint warns the two can
69
+ // report the same import twice); kept for the rule's autofix.
68
70
  "@typescript-eslint/consistent-type-imports": "error",
69
71
  "@typescript-eslint/no-unused-vars": [
70
72
  "error",
@@ -19,9 +19,10 @@ export default defineConfig({
19
19
  // barrel pattern): a fresh scaffold's index.ts IS its real content,
20
20
  // so excluding it would make this gate measure nothing at all.
21
21
  exclude: ["**/*.d.ts"],
22
- // json emits coverage-final.json: the v8 text table hides files that
23
- // are 100% on every metric, so the JSON is the authoritative
24
- // per-file record when investigating a suspected gap.
22
+ // json emits coverage-final.json: the text table can hide files that
23
+ // are 100% on every metric (Vitest sets `skipFull` when it detects an
24
+ // AI coding agent), so the JSON is the authoritative per-file record
25
+ // when investigating a suspected gap.
25
26
  reporter: ["text", "html", "json"],
26
27
  thresholds: {
27
28
  // Scaffold-appropriate starting floor. Raise these to the project's
@@ -105,9 +105,12 @@ already enumerate groups rather than individual steps.
105
105
 
106
106
  | Pack | Contents |
107
107
  | ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
108
- | `harness-extras` | Claude Code session ergonomics: the compaction-handoff hook pair, a read-only Bash guard, and a five-row status line (session, model, context, quota, work) plus a per-subagent row renderer. The only pack that sets top-level settings keys (`statusLine`, `subagentStatusLine`). |
108
+ | `harness-extras` | Claude Code session hooks: the compaction-handoff hook pair (`PreCompact`, `SessionStart`) and a read-only Bash guard for review spokes. Registers hooks only; no top-level settings keys. |
109
+ | `statusline` | A five-row `statusLine` (session, model, context, quota, work) plus a per-subagent row renderer. The only pack that sets top-level settings keys (`statusLine`, `subagentStatusLine`); a project that already defines either must decide on it, see its `adoptNotes`. |
109
110
  | `quality` | Two language-level review aids: a per-file size ratchet gate (`check-file-budget.mjs`, a `build`-group verify step) and a read-only `type-design-analyzer` agent. No hooks, no settings. Adopt-capable and recommended for every project kind. |
110
- | `github` | GitHub-hosted collaboration: Anthropic's official Claude Code GitHub Action wired for `@claude` mention-mode, a second Action that posts an automated Claude review comment on every PR, and three `gh`-CLI skills — `reviewing-dependabot-prs`, `triaging-scan-alerts`, `watching-pr-checks`. No hooks, no gate. The two workflows need an auth secret this pack cannot create; see its `adoptNotes`. |
111
- | `publishing` | A release pipeline: `release.yml` (changesets version-PR / staged, provenance-attested npm publish via trusted publishing), `check-publish-version.mjs`, `check-dts-deps.mjs`, `check-license-headers.mjs` and a `REUSE.toml` template. Fresh mode only; see "The wiring contract" above and its `adoptNotes`. |
111
+ | `claude-github` | Anthropic's official Claude Code GitHub Action: one workflow wired for `@claude` mention-mode and a second that posts an automated Claude review comment on every PR. A pure file drop: no hooks, no gate. The workflows need an auth secret this pack cannot create and the review spends API credits per PR; see its `adoptNotes`. |
112
+ | `github` | GitHub-hosted collaboration through the `gh` CLI: three skills — `reviewing-dependabot-prs`, `triaging-scan-alerts`, `watching-pr-checks`. No hooks, no gate, no workflows. `triaging-scan-alerts` needs code scanning turned on; see its `adoptNotes`. |
113
+ | `publishing` | A release pipeline: `release.yml` (changesets version-PR / staged, provenance-attested npm publish via trusted publishing), `check-publish-version.mjs` and `check-dts-deps.mjs`. Fresh mode only; see "The wiring contract" above and its `adoptNotes`. |
114
+ | `licensing` | An SPDX/REUSE license-header gate: `check-license-headers.mjs` (a `lint`-group verify step) and a `REUSE.toml` template. Fresh mode only: the one-time backfill (`setupSteps`) fails an established project's `pnpm verify` until it has touched every file; see its `adoptNotes`. |
112
115
  | `supply-chain` | Secret scanning (`gitleaks.yml` with a `.gitleaks.toml`) and an OpenSSF Scorecard run (`scorecard.yml`) for any project on GitHub, published or not. A pure file drop of two workflows (read-only apart from the Scorecard job's `security-events` and `id-token` write): no hooks, no settings, no scripts, no gate. Adopt-capable and recommended for every project kind. |
113
116
  | `worktrees` | Enforces that all src/tests development happens inside an isolated git worktree, on any branch, for any caller: a `working-in-worktrees` skill (start/status/sync/finish/fan-out), a `SessionStart` hook that installs dependencies into a freshly created worktree, a `PreToolUse` guard stricter than the baseline's own `guard-branch-isolation.mjs`/`guard-hub-src-writes.mjs`, a `repair-core-bare.mjs` hook (`SessionStart`, and `PostToolUse` after `EnterWorktree`/`ExitWorktree`/`Agent`) that resets the `core.bare = true` Claude Code's worktree tools are reported to leave in a normal repo's shared config, and a `.worktreeinclude` copying `.env`/`.env.local`/`.env.*.local` into every worktree Claude Code creates. Changes the day-to-day workflow, not just a nicety — see its `adoptNotes`. |
@@ -86,7 +86,7 @@ jobs:
86
86
  fetch-depth: 0
87
87
 
88
88
  - id: claude
89
- uses: anthropics/claude-code-action@6fed3ca145920b639991cb756090506e1bcaf515 # v1.0.245
89
+ uses: anthropics/claude-code-action@5eb090f3c3afd701a0672cd32981197ca27688f6 # v1.0.253
90
90
  with:
91
91
  anthropic_api_key: ${{ secrets.ANTHROPIC_API_KEY }}
92
92
  # Swap in CLAUDE_CODE_OAUTH_TOKEN or the Workload Identity
@@ -67,7 +67,7 @@ jobs:
67
67
  fetch-depth: 1
68
68
 
69
69
  - name: Run Claude Code
70
- uses: anthropics/claude-code-action@6fed3ca145920b639991cb756090506e1bcaf515 # v1.0.245
70
+ uses: anthropics/claude-code-action@5eb090f3c3afd701a0672cd32981197ca27688f6 # v1.0.253
71
71
  with:
72
72
  # Option 1 (active default): a stored API key. Install the Claude
73
73
  # GitHub App (https://github.com/apps/claude) and add
@@ -0,0 +1,19 @@
1
+ {
2
+ "schemaVersion": 1,
3
+ "name": "claude-github",
4
+ "description": "Anthropic's official Claude Code GitHub Action (anthropics/claude-code-action) on top of the baseline, as two workflows: claude.yml, wired for @claude mention-mode, and claude-pr-review.yml, which posts an automated Claude review comment on every pull request. Pure file drop: no hooks, no settings, no scripts, no verify step. Both need an auth secret this pack cannot create, and the review workflow spends API credits on every PR, so it is a separate pack from `github` rather than a default.",
5
+ "modes": ["fresh", "adopt"],
6
+ "budget": {
7
+ "agents": 0,
8
+ "skills": 0,
9
+ "hooks": 0,
10
+ "workflows": 2,
11
+ "scripts": 0
12
+ },
13
+ "wiring": {
14
+ "settings": {},
15
+ "packageScripts": {},
16
+ "verifySteps": []
17
+ },
18
+ "adoptNotes": "claude.yml only reacts to @claude mentions from a repository owner, member or collaborator (author_association), so an outside commenter on a public repository cannot start a job that holds contents: write and your auth secret. It also commits through the GitHub API (use_commit_signing) so a signed-commits ruleset accepts its work. claude-pr-review.yml runs on every non-bot, non-fork pull request, so it spends API credits on each push -- delete it and keep claude.yml if you only want on-demand help. Both workflows need one thing this pack cannot verify or install: an ANTHROPIC_API_KEY repository secret (the active default), or a CLAUDE_CODE_OAUTH_TOKEN from a Claude subscription, or a Workload Identity Federation setup in the Anthropic Console -- see the comments in .github/workflows/claude.yml for all three. An adopted project that already has its own .github/workflows/claude.yml or claude-pr-review.yml surfaces as an ordinary file conflict -- keep the existing one if it's already configured for that project's own auth choice, since this pack's version would just overwrite a working setup with its own generic defaults. The two workflows ship without an SPDX header on purpose (a header would carry a project-name token that stays literal in an adopt-mode install); a project that also installs `licensing` gets the headers from that pack's `check-license-headers.mjs --fix` backfill. These workflows used to ship inside the `github` pack; a project that installed that pack before the split keeps its files untouched."
19
+ }
@@ -1,13 +1,13 @@
1
1
  {
2
2
  "schemaVersion": 1,
3
3
  "name": "github",
4
- "description": "GitHub-hosted collaboration on top of the baseline: Anthropic's official Claude Code GitHub Action (anthropics/claude-code-action) wired for @claude mention-mode, a second Action that posts an automated Claude review comment on every PR, and three gh-CLI skills -- reviewing-dependabot-prs (classify and batch-merge open Dependabot PRs), triaging-scan-alerts (triage open code-scanning alerts to file:line, report only), and watching-pr-checks (check a PR's CI status and hand off to a project's own /triaging-ci on a real failure).",
4
+ "description": "GitHub-hosted collaboration skills on top of the baseline, all driven by the gh CLI: reviewing-dependabot-prs (classify and batch-merge open Dependabot PRs), triaging-scan-alerts (triage open code-scanning alerts to file:line, report only), and watching-pr-checks (check a PR's CI status and hand off to a project's own /triaging-ci on a real failure). No hooks, no gate, no workflows: the Claude Code GitHub Action workflows are the separate `claude-github` pack.",
5
5
  "modes": ["fresh", "adopt"],
6
6
  "budget": {
7
7
  "agents": 0,
8
8
  "skills": 3,
9
9
  "hooks": 0,
10
- "workflows": 2,
10
+ "workflows": 0,
11
11
  "scripts": 0
12
12
  },
13
13
  "wiring": {
@@ -15,5 +15,5 @@
15
15
  "packageScripts": {},
16
16
  "verifySteps": []
17
17
  },
18
- "adoptNotes": "claude.yml only reacts to @claude mentions from a repository owner, member or collaborator (author_association), so an outside commenter on a public repository cannot start a job that holds contents: write and your auth secret. It also commits through the GitHub API (use_commit_signing) so a signed-commits ruleset accepts its work. Both workflows need one thing this pack cannot verify or install: an ANTHROPIC_API_KEY repository secret (the active default), or a CLAUDE_CODE_OAUTH_TOKEN from a Claude subscription, or a Workload Identity Federation setup in the Anthropic Console -- see the comments in .github/workflows/claude.yml for all three. An adopted project that already has its own .github/workflows/claude.yml or claude-pr-review.yml surfaces as an ordinary file conflict -- keep the existing one if it's already configured for that project's own auth choice, since this pack's version would just overwrite a working setup with its own generic defaults. triaging-scan-alerts reads code-scanning alerts, and nothing in the baseline or this pack produces them: turn on GitHub's CodeQL default setup for the repository (Settings > Code security), or install the `supply-chain` pack, whose scorecard.yml uploads SARIF, otherwise the skill's first call returns \"code scanning is not enabled\" and reports nothing. The other two skills have no such dependency: they install anywhere a .claude/ directory exists and need only `gh` CLI auth (already required by every other gh-CLI-based skill this baseline ships). watching-pr-checks' handoff to /triaging-ci degrades gracefully if the project doesn't have that skill installed -- it still reports the real failure, just without the handoff."
18
+ "adoptNotes": "The three skills need only `gh` CLI auth (already required by every other gh-CLI-based skill this baseline ships) and install anywhere a .claude/ directory exists. triaging-scan-alerts reads code-scanning alerts, and nothing in the baseline or this pack produces them: turn on GitHub's CodeQL default setup for the repository (Settings > Code security), or install the `supply-chain` pack, whose scorecard.yml uploads SARIF, otherwise the skill's first call returns \"code scanning is not enabled\" and reports nothing. watching-pr-checks' handoff to /triaging-ci degrades gracefully if the project doesn't have that skill installed -- it still reports the real failure, just without the handoff. The Claude Code GitHub Action workflows (claude.yml, claude-pr-review.yml) used to ship in this pack and are now the separate `claude-github` pack; a project that installed this pack before the split keeps those files untouched."
19
19
  }
@@ -1,12 +1,12 @@
1
1
  {
2
2
  "schemaVersion": 1,
3
3
  "name": "harness-extras",
4
- "description": "Adds Claude Code session ergonomics on top of the baseline: a pair of hooks that carry a work summary across a context-window compaction, a hook that blocks non-read-only Bash commands from read-only review spokes, and a five-row statusLine (plus a per-subagent row renderer) -- the only documented surface carrying live context-window pressure.",
4
+ "description": "Adds Claude Code session ergonomics on top of the baseline: a pair of hooks that carry a work summary across a context-window compaction, and a hook that blocks non-read-only Bash commands from read-only review spokes. The status line is the separate `statusline` pack.",
5
5
  "modes": ["fresh", "adopt"],
6
6
  "budget": {
7
7
  "agents": 0,
8
8
  "skills": 0,
9
- "hooks": 6,
9
+ "hooks": 3,
10
10
  "workflows": 0,
11
11
  "scripts": 0
12
12
  },
@@ -51,20 +51,8 @@
51
51
  }
52
52
  ]
53
53
  },
54
- "settingsTopLevel": {
55
- "statusLine": {
56
- "type": "command",
57
- "command": "node \"$CLAUDE_PROJECT_DIR/.claude/hooks/statusline.mjs\"",
58
- "refreshInterval": 30,
59
- "padding": 1
60
- },
61
- "subagentStatusLine": {
62
- "type": "command",
63
- "command": "node \"$CLAUDE_PROJECT_DIR/.claude/hooks/subagent-statusline.mjs\""
64
- }
65
- },
66
54
  "packageScripts": {},
67
55
  "verifySteps": []
68
56
  },
69
- "adoptNotes": "The statusline and compaction hooks install anywhere a .claude/ directory exists, but guard-readonly-bash.mjs imports ../../bin/lib/agent-roster.mjs (the baseline's list of writer spokes) -- copy that file alongside it in an adopted project, or skip that one hook. The PreCompact hook writes tmp/compact-handoff-<session_id>.json under the git worktree the session runs in (from the hook payload's cwd, since CLAUDE_PROJECT_DIR does not follow a session into a linked worktree) -- fresh mode's baseline .gitignore already lists tmp/ unconditionally, but an adopted project's own .gitignore was never touched by adopt mode, so add tmp/ there too before this pack's hooks run, or the handoff artifact can end up committed. The statusline scripts read only the stdin payload, .git/HEAD (via node:fs, never a git subprocess) and process.availableMemory()/os.totalmem(), so they install anywhere a .claude/ directory exists -- no dependency on the baseline's file layout. The one real adopt risk is the two top-level settings keys: a project that already defines statusLine or subagentStatusLine collides, and its existing value must be shown and decided on, never overwritten -- if either is already set, skip that key and the three statusline scripts, and install the rest of this pack (the hooks) normally rather than failing the whole install. A statusLine in .claude/settings.local.json (or one passed with --settings) also shadows the project one; the user's own ~/.claude/settings.json ranks lower and does not -- check those before concluding the pack is wired. rate_limits.spend_limit and prompt_cache need Claude Code v2.1.251+, and the per-subagent model/contextWindowSize/effort fields need v2.1.205+/v2.1.214+, and the agentType label a row falls back to when a subagent has no name needs v2.1.293+; each renders only when present, so an older Claude Code degrades to fewer segments rather than breaking. The scripts behave the same on macOS and Linux; the memory segment uses process.availableMemory() (Node 22+), and on macOS with an older Node it is hidden rather than shown from os.freemem(), which undercounts there. Below roughly 40 columns the layout drops segments by priority rather than wrapping. The size-ratchet gate and the type-design-analyzer agent that used to ship here are now the separate `quality` pack."
57
+ "adoptNotes": "The compaction hooks install anywhere a .claude/ directory exists, but guard-readonly-bash.mjs imports ../../bin/lib/agent-roster.mjs (the baseline's list of writer spokes) -- copy that file alongside it in an adopted project, or skip that one hook. The PreCompact hook writes tmp/compact-handoff-<session_id>.json under the git worktree the session runs in (from the hook payload's cwd, since CLAUDE_PROJECT_DIR does not follow a session into a linked worktree) -- fresh mode's baseline .gitignore already lists tmp/ unconditionally, but an adopted project's own .gitignore was never touched by adopt mode, so add tmp/ there too before this pack's hooks run, or the handoff artifact can end up committed. The size-ratchet gate and the type-design-analyzer agent that used to ship here are now the separate `quality` pack. The status line that used to ship here is now the separate `statusline` pack, so installing this one never touches a project's own `statusLine` setting."
70
58
  }
@@ -0,0 +1,31 @@
1
+ {
2
+ "schemaVersion": 1,
3
+ "name": "licensing",
4
+ "description": "An SPDX/REUSE license-header gate for a new project: check-license-headers.mjs (a lint-group verify step that fails on any git-tracked file with neither an SPDX header nor a REUSE.toml annotation, and inserts missing headers with --fix) and a REUSE.toml template. Fresh mode only, because the one-time backfill fails an established project's pnpm verify until it has touched every file. Split from `publishing` so a project can take the header gate without the npm release pipeline, and the other way round.",
5
+ "modes": ["fresh"],
6
+ "budget": {
7
+ "agents": 0,
8
+ "skills": 0,
9
+ "hooks": 0,
10
+ "workflows": 0,
11
+ "scripts": 0
12
+ },
13
+ "wiring": {
14
+ "settings": {},
15
+ "packageScripts": {},
16
+ "verifySteps": [
17
+ {
18
+ "id": "license-headers",
19
+ "group": "lint",
20
+ "name": "Check SPDX license headers",
21
+ "cmd": ["node", "bin/check-license-headers.mjs"]
22
+ }
23
+ ]
24
+ },
25
+ "setupSteps": [
26
+ "git add -A",
27
+ "node bin/check-license-headers.mjs --fix",
28
+ "git add -A"
29
+ ],
30
+ "adoptNotes": "REQUIRED right after install, before the first pnpm verify: `git add -A`, then `node bin/check-license-headers.mjs --fix`, then `git add -A` again and commit the result (the gate only checks git-tracked files, so a fresh project with nothing staged passes it vacuously and fails on every file at its first push) -- templates/core has never had a license-header gate before this pack, so every one of its own files (hooks, bin/ scripts, CI workflows, the placeholder src/tests, and the workflows of any other installed pack such as `claude-github` or `supply-chain`) is missing the SPDX header this pack's gate now requires. The shipped REUSE.toml and gate header use NOASSERTION as the license identifier because the pack cannot know the project's license: edit REUSE.toml and the headerLines() function in bin/check-license-headers.mjs once it is decided (a LICENSE file, if one exists, is the source of truth). In a hand-copied install the CLI does not substitute tokens, so replace every literal `__PROJECT_NAME__` in the copied files (REUSE.toml, the SPDX headers) with the project's real name first. Fresh mode only: the backfill fails an established project's pnpm verify until it has touched every file, so it is staged like every other pack (at .groundwork/packs/licensing/) but never auto-installed in adopt mode -- if an adopted project wants it, copy .groundwork/packs/licensing/files/ by hand, dropping the `.staged` suffix from every name (or copy templates/packs/licensing/files/ from this repo directly) and run the backfill above. The license-headers verify step is a lint-group gate and installs anywhere a bin/verify.mjs-shaped gate runner exists; wire it into the project's real one if it differs. REUSE.toml also carries globs for `.gitleaks.toml` and `.worktreeinclude`, the files the `supply-chain` and `worktrees` packs add, so the gate's uncovered-file check passes whether or not those packs are installed. This gate used to ship in the `publishing` pack; a project that installed that pack before the split keeps its files untouched."
31
+ }
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "schemaVersion": 1,
3
3
  "name": "publishing",
4
- "description": "A release pipeline for a project that ships an npm package. Release half: release.yml (changesets version-PR / staged, provenance-attested npm publish via trusted publishing OIDC, including a direct check-publish-version.mjs call right before packing), .changeset/config.json + README.md, the pnpm-to-staged-publish shim (libnpmpublish, not the npm CLI), and one build-group gate (check-dts-deps.mjs). Also ships check-license-headers.mjs and a REUSE.toml template as a lint-group gate (secret scanning and Scorecard are the separate `supply-chain` pack). Not adopt-capable: the release flow assumes decisions (registry access, npm trusted-publisher setup, a GitHub App for the version PR) too project-specific to install blind -- see adoptNotes.",
4
+ "description": "A release pipeline for a project that ships an npm package. Release half: release.yml (changesets version-PR / staged, provenance-attested npm publish via trusted publishing OIDC, including a direct check-publish-version.mjs call right before packing), .changeset/config.json + README.md, the pnpm-to-staged-publish shim (libnpmpublish, not the npm CLI), and one build-group gate (check-dts-deps.mjs). The SPDX license-header gate is the separate `licensing` pack, and secret scanning and Scorecard are the separate `supply-chain` pack. Not adopt-capable: the release flow assumes decisions (registry access, npm trusted-publisher setup, a GitHub App for the version PR) too project-specific to install blind -- see adoptNotes.",
5
5
  "modes": ["fresh"],
6
6
  "budget": {
7
7
  "agents": 0,
@@ -22,20 +22,9 @@
22
22
  "group": "build",
23
23
  "name": "Check .d.ts dependency declarations",
24
24
  "cmd": ["node", "bin/check-dts-deps.mjs"]
25
- },
26
- {
27
- "id": "license-headers",
28
- "group": "lint",
29
- "name": "Check SPDX license headers",
30
- "cmd": ["node", "bin/check-license-headers.mjs"]
31
25
  }
32
26
  ]
33
27
  },
34
- "setupSteps": [
35
- "pnpm add -D @changesets/cli",
36
- "git add -A",
37
- "node bin/check-license-headers.mjs --fix",
38
- "git add -A"
39
- ],
40
- "adoptNotes": "REQUIRED right after install, before the first pnpm verify -- two one-time steps, or pnpm verify fails immediately, not from a real regression but from two setup gaps this pack cannot close itself: (1) `pnpm add -D @changesets/cli` -- the wiring contract only extends package.json's scripts, never its dependencies (see templates/packs/README.md), so the wired `changeset`/`version:packages` scripts reference a binary knip's own unlisted-binaries check correctly flags as missing until this is installed; (2) `git add -A`, then `node bin/check-license-headers.mjs --fix`, then `git add -A` again and commit the result (the gate only checks git-tracked files, so a fresh project with nothing staged passes it vacuously and fails on every file at its first push) -- templates/core has never had a license-header gate before this pack, so every one of its own files (hooks, bin/ scripts, CI workflows, the placeholder src/tests, and the workflows of any other installed pack such as `github` or `supply-chain`) is missing the SPDX header this pack's own license-headers gate now requires, the same one-time cost this repo itself paid when it adopted the same gate (see CLAUDE.md's Definition of Done). In a hand-copied install the CLI does not substitute tokens, so replace every literal `__PROJECT_NAME__` in the copied files (REUSE.toml, the SPDX headers in bin/, the workflows) with the project's real name first. Fresh mode only: a release pipeline encodes decisions (whether the package is actually public, which registry, whether a GitHub App backs the version PR) this pack cannot survey or guess at safely, so it is staged like every other pack (at .groundwork/packs/publishing/) but never auto-installed in adopt mode -- if an adopted project wants it, copy .groundwork/packs/publishing/files/ by hand, dropping the `.staged` suffix from every name (or copy templates/packs/publishing/files/ from this repo directly) and work through the one-time setup below. check-publish-version.mjs is deliberately NOT one of this pack's wired verify steps: with changesets, package.json's version on the release branch is always the one just published between releases, so a version-already-published check running on every ordinary push (via pnpm verify / pre-push) would fail every single time until the next version PR lands -- it is instead invoked directly as a step inside release.yml's pack job, the one place \"is this version about to collide with an already-published one\" is actually the right question to ask. The two remaining verify-group gates (dts-deps, license-headers) install anywhere a bin/verify.mjs-shaped gate runner exists; wire them into the project's real one if it differs. Both check-publish-version.mjs and check-dts-deps.mjs no-op (checked: 0, not a failure) on a package.json with `private: true` (the baseline's own default) or (check-dts-deps.mjs only) with no dist/**/*.d.ts yet -- flip `private` to `false` (and set a real `name`) once the package is actually meant to publish. The pack assumes a public package: `.changeset/config.json` ships `access: public`, npm refuses `restricted` for an unscoped name, and the publisher always requests provenance, which npm ties to a public source repository -- a private scoped package needs `access: restricted` and the `provenance: true` option removed from the libnpmpublish call in bin/npm-stage-publish.mjs. `@changesets/changelog-github` (and its `repo` option) is a drop-in swap for `.changeset/config.json`'s plain default changelog generator, once this project's GitHub repository is known, for changelog entries that link back to the originating PR/commit -- see `.changeset/README.md`. One-time setup before release.yml's first real run (distinct from the two pnpm-verify prerequisites above), distilled from this repo's own .claude/rules/releases.md: (1) npm cannot configure a trusted publisher for a package that doesn't exist yet -- publish once by hand with a temporary token first; (2) the trusted publisher is bound to the exact workflow filename `release.yml` -- renaming or moving it breaks publishing until reconfigured on npmjs.com; (3) set the trusted publisher's allowed actions to staged-only (staged publishing only, npm's own default and recommendation since 2026-09-03) -- a maintainer runs `npm stage approve <id>` (2FA, never automatable) before a version actually installs; (4) create an `npm-publish` GitHub environment (via `gh api`, not committed as JSON) with a required reviewer, restricted to the release branch, so the `publish` job itself pauses before the git tag/Release/stage-publish exist; (5) if branch protection requires status checks on every PR with no bypass actor, the version-PR job needs a GitHub App installation token (`APP_CLIENT_ID`/`APP_PRIVATE_KEY` repo secrets), not the default GITHUB_TOKEN, or those checks never trigger and the PR can never merge -- if this project has no such rule, `github-token: ${{ secrets.GITHUB_TOKEN }}` is enough and the app-token step can be dropped. release.yml pins `tools/release/` (the exact libnpmpublish and npm-registry-fetch versions the publish job installs; it stages through them, not the npm CLI, because the CLI's bundled dependencies cannot be patched by an override) but the baseline's dependabot.yml only covers the github-actions ecosystem, so that pin goes stale unless you add an `npm` entry with `directory: \"/tools/release\"` to the project's own .github/dependabot.yml. The staged id to approve is printed by the workflow's \"Show the staged ids to approve\" step (and annotated on the run), and the `npm stage approve <id>` command is also written to the job summary. The secret-scanning and Scorecard workflows that used to ship here are now the separate `supply-chain` pack."
28
+ "setupSteps": ["pnpm add -D @changesets/cli"],
29
+ "adoptNotes": "REQUIRED right after install, before the first pnpm verify -- one step, or pnpm verify fails immediately, not from a real regression but from a setup gap this pack cannot close itself: `pnpm add -D @changesets/cli` -- the wiring contract only extends package.json's scripts, never its dependencies (see templates/packs/README.md), so the wired `changeset`/`version:packages` scripts reference a binary knip's own unlisted-binaries check correctly flags as missing until this is installed. In a hand-copied install the CLI does not substitute tokens, so replace every literal `__PROJECT_NAME__` in the copied files (the SPDX headers in bin/, the workflows) with the project's real name first. Fresh mode only: a release pipeline encodes decisions (whether the package is actually public, which registry, whether a GitHub App backs the version PR) this pack cannot survey or guess at safely, so it is staged like every other pack (at .groundwork/packs/publishing/) but never auto-installed in adopt mode -- if an adopted project wants it, copy .groundwork/packs/publishing/files/ by hand, dropping the `.staged` suffix from every name (or copy templates/packs/publishing/files/ from this repo directly) and work through the one-time setup below. check-publish-version.mjs is deliberately NOT one of this pack's wired verify steps: with changesets, package.json's version on the release branch is always the one just published between releases, so a version-already-published check running on every ordinary push (via pnpm verify / pre-push) would fail every single time until the next version PR lands -- it is instead invoked directly as a step inside release.yml's pack job, the one place \"is this version about to collide with an already-published one\" is actually the right question to ask. The remaining verify-group gate (dts-deps) installs anywhere a bin/verify.mjs-shaped gate runner exists; wire it into the project's real one if it differs. Both check-publish-version.mjs and check-dts-deps.mjs no-op (checked: 0, not a failure) on a package.json with `private: true` (the baseline's own default) or (check-dts-deps.mjs only) with no dist/**/*.d.ts yet -- flip `private` to `false` (and set a real `name`) once the package is actually meant to publish. The pack assumes a public package: `.changeset/config.json` ships `access: public`, npm refuses `restricted` for an unscoped name, and the publisher always requests provenance, which npm ties to a public source repository -- a private scoped package needs `access: restricted` and the `provenance: true` option removed from the libnpmpublish call in bin/npm-stage-publish.mjs. `@changesets/changelog-github` (and its `repo` option) is a drop-in swap for `.changeset/config.json`'s plain default changelog generator, once this project's GitHub repository is known, for changelog entries that link back to the originating PR/commit -- see `.changeset/README.md`. One-time setup before release.yml's first real run (distinct from the pnpm-verify prerequisite above), distilled from this repo's own .claude/rules/releases.md: (1) npm cannot configure a trusted publisher for a package that doesn't exist yet -- publish once by hand with a temporary token first; (2) the trusted publisher is bound to the exact workflow filename `release.yml` -- renaming or moving it breaks publishing until reconfigured on npmjs.com; (3) set the trusted publisher's allowed actions to staged-only (staged publishing only, npm's own default and recommendation since 2026-09-03) -- a maintainer runs `npm stage approve <id>` (2FA, never automatable) before a version actually installs; (4) create an `npm-publish` GitHub environment (via `gh api`, not committed as JSON) with a required reviewer, restricted to the release branch, so the `publish` job itself pauses before the git tag/Release/stage-publish exist; (5) if branch protection requires status checks on every PR with no bypass actor, the version-PR job needs a GitHub App installation token (`APP_CLIENT_ID`/`APP_PRIVATE_KEY` repo secrets), not the default GITHUB_TOKEN, or those checks never trigger and the PR can never merge -- if this project has no such rule, `github-token: ${{ secrets.GITHUB_TOKEN }}` is enough and the app-token step can be dropped. release.yml pins `tools/release/` (the exact libnpmpublish and npm-registry-fetch versions the publish job installs; it stages through them, not the npm CLI, because the CLI's bundled dependencies cannot be patched by an override) but the baseline's dependabot.yml only covers the github-actions ecosystem, so that pin goes stale unless you add an `npm` entry with `directory: \"/tools/release\"` to the project's own .github/dependabot.yml. The staged id to approve is printed by the workflow's \"Show the staged ids to approve\" step (and annotated on the run), and the `npm stage approve <id>` command is also written to the job summary. The secret-scanning and Scorecard workflows that used to ship here are now the separate `supply-chain` pack. The SPDX/REUSE license-header gate that used to ship here is now the separate fresh-only `licensing` pack: this pack's own scripts and release.yml still carry SPDX headers with a NOASSERTION license identifier, which is harmless without that gate but is what it would check."
41
30
  }
@@ -0,0 +1,31 @@
1
+ {
2
+ "schemaVersion": 1,
3
+ "name": "statusline",
4
+ "description": "A five-row Claude Code statusLine (session, model, context, quota, work) plus a per-subagent row renderer -- the only documented surface carrying live context-window pressure. Three scripts and two top-level settings keys (`statusLine`, `subagentStatusLine`): no hook registrations, no verify step. Split from `harness-extras` so a project that already has its own status line can take that pack's hooks without a collision.",
5
+ "modes": ["fresh", "adopt"],
6
+ "budget": {
7
+ "agents": 0,
8
+ "skills": 0,
9
+ "hooks": 3,
10
+ "workflows": 0,
11
+ "scripts": 0
12
+ },
13
+ "wiring": {
14
+ "settings": {},
15
+ "settingsTopLevel": {
16
+ "statusLine": {
17
+ "type": "command",
18
+ "command": "node \"$CLAUDE_PROJECT_DIR/.claude/hooks/statusline.mjs\"",
19
+ "refreshInterval": 30,
20
+ "padding": 1
21
+ },
22
+ "subagentStatusLine": {
23
+ "type": "command",
24
+ "command": "node \"$CLAUDE_PROJECT_DIR/.claude/hooks/subagent-statusline.mjs\""
25
+ }
26
+ },
27
+ "packageScripts": {},
28
+ "verifySteps": []
29
+ },
30
+ "adoptNotes": "The statusline scripts read only the stdin payload, .git/HEAD (via node:fs, never a git subprocess) and process.availableMemory()/os.totalmem(), so they install anywhere a .claude/ directory exists -- no dependency on the baseline's file layout. The one real adopt risk is the two top-level settings keys: a project that already defines statusLine or subagentStatusLine collides, and its existing value must be shown and decided on, never overwritten -- if either is already set, skip that key and the script that renders it rather than failing the whole install; if both are set there is nothing left to install, so skip the pack. A statusLine in .claude/settings.local.json (or one passed with --settings) also shadows the project one; the user's own ~/.claude/settings.json ranks lower and does not -- check those before concluding the pack is wired. rate_limits.spend_limit and prompt_cache need Claude Code v2.1.251+, and the per-subagent model/contextWindowSize/effort fields need v2.1.205+/v2.1.214+, and the agentType label a row falls back to when a subagent has no name needs v2.1.293+; each renders only when present, so an older Claude Code degrades to fewer segments rather than breaking. The scripts behave the same on macOS and Linux; the memory segment uses process.availableMemory() (Node 22+), and on macOS with an older Node it is hidden rather than shown from os.freemem(), which undercounts there. Below roughly 40 columns the layout drops segments by priority rather than wrapping. These scripts used to ship inside the `harness-extras` pack; a project that installed that pack before the split keeps its files untouched."
31
+ }
@@ -15,5 +15,5 @@
15
15
  "packageScripts": {},
16
16
  "verifySteps": []
17
17
  },
18
- "adoptNotes": "An adopted project that already has its own .github/workflows/gitleaks.yml or scorecard.yml surfaces as an ordinary file conflict -- keep the existing one if it already works, since this pack's version would just replace a configured setup with its generic defaults. gitleaks.yml needs a free GITLEAKS_LICENSE repo secret for an organization-owned repository (gitleaks-action's own requirement); scorecard.yml assumes a public repository (`publish_results: true` needs one) -- set it `false` for a private repo. gitleaks.yml is read-only and scorecard.yml holds security-events: write and id-token: write in its analysis job; neither is a required check by default, so add either to branch protection once it has a clean run. The SPDX/REUSE license-header gate is not part of this pack: it lives in the fresh-only `publishing` pack, because it fails an established project's `pnpm verify` until a one-time backfill touches every file. The two workflows ship without an SPDX header on purpose (a header would carry a project-name token that stays literal in an adopt-mode install); a project that also installs `publishing` gets the headers from that pack's `check-license-headers.mjs --fix` backfill."
18
+ "adoptNotes": "An adopted project that already has its own .github/workflows/gitleaks.yml or scorecard.yml surfaces as an ordinary file conflict -- keep the existing one if it already works, since this pack's version would just replace a configured setup with its generic defaults. gitleaks.yml needs a free GITLEAKS_LICENSE repo secret for an organization-owned repository (gitleaks-action's own requirement); scorecard.yml assumes a public repository (`publish_results: true` needs one) -- set it `false` for a private repo. gitleaks.yml is read-only and scorecard.yml holds security-events: write and id-token: write in its analysis job; neither is a required check by default, so add either to branch protection once it has a clean run. The SPDX/REUSE license-header gate is not part of this pack: it lives in the fresh-only `licensing` pack, because it fails an established project's `pnpm verify` until a one-time backfill touches every file. The two workflows ship without an SPDX header on purpose (a header would carry a project-name token that stays literal in an adopt-mode install); a project that also installs `licensing` gets the headers from that pack's `check-license-headers.mjs --fix` backfill."
19
19
  }
@@ -39,6 +39,7 @@
39
39
  {
40
40
  "type": "command",
41
41
  "command": "node \"$CLAUDE_PROJECT_DIR/.claude/hooks/guard-worktree-only.mjs\"",
42
+ "onFailure": "block",
42
43
  "timeout": 30
43
44
  }
44
45
  ]
@@ -60,5 +61,5 @@
60
61
  "packageScripts": {},
61
62
  "verifySteps": []
62
63
  },
63
- "adoptNotes": "ensure-worktree-deps.mjs runs `pnpm install --frozen-lockfile --prefer-offline` automatically the first time a session starts inside a freshly created worktree with no node_modules yet -- this runs every dependency's lifecycle scripts the same way any `pnpm install` does, limited only by whatever `allowBuilds`/`onlyBuiltDependencies` policy the project's own pnpm config already sets (see typescript-guidance's pnpm supply-chain coverage if none is set). If node_modules in the worktree is a symlink (the usual result of `worktree.symlinkDirectories: [\"node_modules\"]` in .claude/settings.json, which this hook never reads), the hook detects the symlink and skips -- installing there would mutate the main checkout's own node_modules through the symlink. guard-worktree-only.mjs blocks every src/tests write outside a linked worktree under .claude/worktrees/, on any branch, for any caller (hub or writer spoke) -- this is stricter than guard-branch-isolation.mjs (which only blocks on main) and guard-hub-src-writes.mjs (which blocks the hub and any non-writer subagent); confirm the team actually wants every change routed through a worktree before installing this pack, since it changes the day-to-day workflow, not just adds a nicety. `.claude/worktrees/` should already be gitignored (the baseline's own .gitignore covers this); if adopting into a project with a hand-rolled .gitignore that doesn't, add it before installing. repair-core-bare.mjs resets `core.bare = true` to `false` in a non-bare repo's shared .git/config at session start and after EnterWorktree/ExitWorktree/Agent calls: Claude Code's worktree tools have been reported to leave it set (anthropics/claude-code#58345, #69802: user bug reports, both since closed as not planned, so whether it still reproduces is unconfirmed), which breaks `git status` in the main checkout. It only acts on a provably non-bare repository (its common git dir is named .git and contains an index), never on a genuine bare repo -- including a bare clone stored in a directory called .git -- and reports a failed repair on stdout rather than silently."
64
+ "adoptNotes": "ensure-worktree-deps.mjs runs `pnpm install --frozen-lockfile --prefer-offline` automatically the first time a session starts inside a freshly created worktree with no node_modules yet -- this runs every dependency's lifecycle scripts the same way any `pnpm install` does, limited only by whatever `allowBuilds`/`onlyBuiltDependencies` policy the project's own pnpm config already sets (see typescript-guidance's pnpm supply-chain coverage if none is set). If node_modules in the worktree is a symlink (the usual result of `worktree.symlinkDirectories: [\"node_modules\"]` in .claude/settings.json, which this hook never reads), the hook detects the symlink and skips -- installing there would mutate the main checkout's own node_modules through the symlink. guard-worktree-only.mjs is registered with `\"onFailure\": \"block\"` (Claude Code 2.1.295 or later; an older version ignores the key), so a hook that cannot start, times out, exits with a code other than 0 or 2 or prints invalid JSON blocks the write instead of letting it through. A project that installed this pack before it carried that key has the same registration without it: /customize adds the missing key when it installs the pack again, and otherwise add `\"onFailure\": \"block\"` to that registration in .claude/settings.json by hand. It blocks every src/tests write outside a linked worktree under .claude/worktrees/, on any branch, for any caller (hub or writer spoke) -- this is stricter than guard-branch-isolation.mjs (which only blocks on main) and guard-hub-src-writes.mjs (which blocks the hub and any non-writer subagent); confirm the team actually wants every change routed through a worktree before installing this pack, since it changes the day-to-day workflow, not just adds a nicety. `.claude/worktrees/` should already be gitignored (the baseline's own .gitignore covers this); if adopting into a project with a hand-rolled .gitignore that doesn't, add it before installing. repair-core-bare.mjs resets `core.bare = true` to `false` in a non-bare repo's shared .git/config at session start and after EnterWorktree/ExitWorktree/Agent calls: Claude Code's worktree tools have been reported to leave it set (anthropics/claude-code#58345, #69802: user bug reports, both since closed as not planned, so whether it still reproduces is unconfirmed), which breaks `git status` in the main checkout. It only acts on a provably non-bare repository (its common git dir is named .git and contains an index), never on a genuine bare repo -- including a bare clone stored in a directory called .git -- and reports a failed repair on stdout rather than silently."
64
65
  }