SLOPSHOPPER

diff-review

After every Edit, Write, NotebookEdit or sed -i, a docked pane shows that file's uncommitted hunk with [keep] and [revert] buttons; a press dismisses or runs…

newpanespinnerguardcommandtoast
★ 1v0.1.0MITupdated 2026-09-21yash-gadodia/claude-mods/diff-review
A shopper browsing a rack in a slop shop
Preview · a replayed session in a sandbox
claude · ~/work/app · diff-review
│ ┃ diff-review ✕ › fix the failing auth test and add an audit log call │ ┃ │ ┃ src/cache.ts [ keep ] [ revert ] ⏺ Read(src/auth.ts) │ ┃ 0 diff --git a/src/auth.ts b/src/auth.ts ⎿ Read 6 lines │ ┃ 1 --- a/src/auth.ts ⏺ Update(src/auth.ts) │ ┃ 1 +++ b/src/auth.ts ⎿ Added 2 lines, removed 1 line │ ┃ 3 export async function refresh(token: s ⏺ Bash(bun test) │ ┃ 4 const claims = await verify(token) ⎿ 3 pass, 1 fail │ ┃ 5 - if (!claims) throw new Error('invali │ ┃ 5 + if (!claims || claims.exp < Date.now ● Done. refresh now rejects expired claims and logs an audit event. │ ┃ 6 + await audit('refresh', claims.sub) │ ┃ 7 return issue(claims.sub) ✻ Worked for 42s · done 4:20 PM │ ┃ │ ┃ src/audit.ts [ keep ] [ revert ] › /diff-review │ ┃ 0 diff --git a/src/auth.ts b/src/auth.ts ⎿ diff-review: diff-review is on; pane closed; 3 file(s) this turn │ ┃ 1 --- a/src/auth.ts │ ┃ 1 +++ b/src/auth.ts │ ┃ 3 export async function refresh(token: s │ ┃ 4 const claims = await verify(token) │ ┃ 5 - if (!claims) throw new Error('invali │ ┃ 5 + if (!claims || claims.exp < Date.now │ ┃ 6 + await audit('refresh', claims.sub) │ ┃ 7 return issue(claims.sub) │ ┃ │ ┃ src/auth.ts [ keep ] [ revert ] │ ┃ 0 diff --git a/src/auth.ts b/src/auth.ts ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── › ? for shortcuts

Draws

Pane · diff-review
src/cache.ts [ keep ] [ revert ] 0 diff --git a/src/auth.ts b/src/auth.ts 1 --- a/src/auth.ts 1 +++ b/src/auth.ts 3 export async function refresh(token: string) { 4 const claims = await verify(token) 5 - if (!claims) throw new Error('invalid token') 5 + if (!claims || claims.exp < Date.now() / 1000) throw 6 + await audit('refresh', claims.sub) 7 return issue(claims.sub) src/audit.ts [ keep ] [ revert ] 0 diff --git a/src/auth.ts b/src/auth.ts 1 --- a/src/auth.ts 1 +++ b/src/auth.ts 3 export async function refresh(token: string) { 4 const claims = await verify(token) 5 - if (!claims) throw new Error('invalid token') 5 + if (!claims || claims.exp < Date.now() / 1000) throw 6 + await audit('refresh', claims.sub) 7 return issue(claims.sub) src/auth.ts [ keep ] [ revert ] 0 diff --git a/src/auth.ts b/src/auth.ts 1 --- a/src/auth.ts 1 +++ b/src/auth.ts 3 export async function refresh(token: string) { 4 const claims = await verify(token) 5 - if (!claims) throw new Error('invalid token') 5 + if (!claims || claims.exp < Date.now() / 1000) throw 6 + await audit('refresh', claims.sub) 7 return issue(claims.sub)
README

claude-mods

Mods that keep an agent honest — plus a few that make the terminal fun.

claude-code · mod · function-hooks · typescript · macos

License: MIT Claude Code — mod test

usage  max  volty  opus-5  5h 34%  7d 12%  ctx 41% 82k  $1.23
scope: 3/4 files
 ▸

<sub>Thirteen mods, each drawing or guarding its own slice of the session. Above: usage-band and scope-guard.</sub>

usage-band, wod-band and wod-timer above the prompt <sub>usage-band, wod-band and wod-timer in a live session.</sub>

Why

Claude Code will tell you a deploy worked because git push exited 0. It will turn a one-line fix into a nine-file refactor and never mention it. Written rules in CLAUDE.md help until the model forgets them, and you find out on the deploy that breaks.

These are the same rules, moved out of prose and into the engine — where they hold whether or not the model remembers.

What a mod is

A mod is a Claude Code plugin whose behaviour lives in a TypeScript hooks module — register(on, options) wiring handlers onto engine events (tool.call, ui.render, turn.complete) rather than markdown the model reads. A mod can deny a tool call, rewrite it in flight, draw above the prompt, or put evidence in front of the model that it cannot argue with.

Every mod here is source you can read in one sitting. None of them phone home: there is no $.http.fetch anywhere in this repo.

Install

Function hooks are behind a flag. Set it first, in your shell profile or settings.json env:

export CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1

Then, in Claude Code:

/plugin marketplace add yash-gadodia/claude-mods
/plugin install scope-guard@claude-mods

Install only what you want — each mod is independent. Update with claude plugin update <name>@claude-mods.

The mods

Discipline

ModWhat it does
scope-guardCounts the distinct files one turn edits. At the threshold it stops and makes the goal get restated, so a small ask cannot quietly become a refactor. /scope sets it.
deploy-verifyAfter a deploy command succeeds, waits for the GitHub Actions run it started, then curls the live URL with cache-busting and puts the verdict in the model's context. A deploy cannot be claimed without evidence.
receiptThe turn footer becomes a receipt: edits, runs and curls, with a warning when edits ran nothing. Destructive commands are never folded into a tool group, a claim of "fixed" with no run puts "unverified claim pending" in the spinner, and Tab suggests running the tests.
diff-reviewA docked pane with each edited file's hunk and keep or revert buttons. Reverting runs git directly; no model turn.
merge-gateDenies gh pr merge, a git merge on trunk, or a push to main unless the latest human message contains the word merge. Ship, push and deploy do not count. /merge-gate toggles it.
mini-offloadRewrites heavy Bash commands (test suites, builds, Docker) to run on a second machine over ssh — syncing the commit there first, because the remote checkout is the real hazard. /mini sets always, ask, or off.

Instruments

ModWhat it does
usage-bandThe 5-hour and 7-day limit windows, this session's context fill and cost, above the prompt. Nudges you to /clear when the window gets expensive.
money-bandLiquid assets, CPF, debt and month-to-date spend, read from a pair of SQLite databases over ssh. Every figure is the database's own; nothing is estimated.
copy-bandClick-to-copy buttons above the prompt for every code block and quoted draft in the last answer, plus a durable stash of older ones. Copying runs pbcopy directly — no model turn.
done-blinkWhen a turn lands, the iTerm2 tab blinks orange every half second until you send the next prompt or three minutes pass, so a finished session is obvious from any other tab. Works inside tmux with no passthrough config: the escape goes to the tmux client's tty. /done-blink 60 sets the ceiling.
chrome-switchSwitches the Claude in Chrome extension between named browser profiles using select_browser, which needs no approval click. /chromep maps them.

Fun

ModWhat it does
wod-bandA pixel-art athlete above the prompt who does a rep every turn. The session is an AMRAP of thrusters, burpees and pull-ups.
wod-timer3, 2, 1, GO in the spinner when you submit, a running gym clock while Claude works, and a whiteboard split in the footer when the turn lands: turn, time, AMRAP total, PR. /wod-timer voice on reads long splits aloud.

Turning them off

Every mod checks one environment variable before doing anything:

CLAUDE_MODS_DISABLE=all            # every mod in this repo becomes a pass-through
CLAUDE_MODS_DISABLE=scope-guard    # just that one
CLAUDE_MODS_DISABLE=wod-band,wod-timer

A disabled mod registers no command and every hook falls straight through to next(e).

Configuration

Mods that touch your machine declare their settings in plugin.json userConfig, so they are editable through /config rather than by hand:

  • scope-guard — /scope <n> sets the file threshold. /scope judge on|off (default on) lets a one-shot Haiku call decide at the threshold whether the next edit is still inside the goal you stated first; a yes raises the ceiling by one for that turn, a no or a failed call falls back to asking. /scope off disables the guard.
  • merge-gate — /merge-gate on|off. "merge x3" or "merge after each" in your message grants that many merges.
  • mini-offload — host (ssh alias, default mini), remotePath (the PATH export prefixed to every offloaded command). Per-repo overrides live at <repo>/.claude/mini-offload.json.
  • money-band — host, networthDb, financeDb. Expects SQLite databases with accounts/balances and transactions tables. efAccount (default UOB One) and efTarget (default 30000) feed the EF 41% footer label.
  • usage-band — sgdRate (default 1.30) for the S$ footer label; /usage-band sgd off hides it.
  • receipt — /receipt on|off|status.
  • done-blink — /done-blink on|off|status|<seconds> (default 180, max 900).
  • diff-review — /diff-review open|close|on|off.
  • deploy-verify — per-repo, at <repo>/.claude/deploy-verify.json: ``json { "url": "https://example.com", "matchFile": "VERSION" } ``
  • chrome-switch — ~/.claude/chrome-browsers.json, mapping labels to deviceIds.

Evidence, not decoration

scope-guard and deploy-verify also write a block into the model's own context (prompt.context), replacing their previous copy rather than accumulating:

# deployVerify
Last live deploy check, 2 minutes ago:
  VERIFIED live: https://example.com served "v3.10.10"
This is the only evidence about the live site in this session. Do not describe the deploy as
verified unless a line above starts with VERIFIED, and do not re-state an older claim over it.

A band above the prompt is for you. A context block is for the model — and it cannot be talked around. Repeated advisories are hashed and suppressed for a cooldown so this costs context once, not once per tool call; verdicts themselves are never throttled, because a verdict is evidence.

Tests

npm install
npm test

npm test typechecks every mod, runs its suite under claude plugin test (the official kit, claude-code/testing, with a mocked clock, store and process table), and checks each mod's footprint: the hooks, $ calls and env reads that claude plugin validate reports, pinned in <mod>/FOOTPRINT. A mod that starts calling $.http.fetch fails the build instead of a README sentence going stale. scripts/footprint.sh --write re-pins after a deliberate change.

The interesting half of deploy-verify's suite is the clean baseline: commands that mention a deploy without being one — echo "git push", grep -r "wrangler deploy", git push --dry-run, a commit message quoting make deploy, a heredoc containing one. A false positive curls a live URL nothing was pushed to and then reports a verdict about it, which is worse than not checking at all.

Design rules

The ones that survived contact with real sessions:

  1. A deny always carries the fix. Blocking without saying what to do instead strands the model in a retry loop. Every refusal here names the next action.
  2. Never block when there is no way through. If the only outcomes are "denied" and "denied again", let it run and say something instead.
  3. A render hook that throws takes the whole mod down with it. Every band wraps its frame in try/catch and falls back to what was there.
  4. Hooks have ten seconds of their own time. next(e) and $ calls are free; $.clock.sleep is not. Past the budget, or on a throw, the engine skips the hook silently unless it declares .catch — so every guard here catches and denies, and slow work belongs on a timer.
  5. Bands yield. e.props.hasSurvey means the engine wants that slot; give it back.

Requirements

Claude Code 2.1.271+ with CLAUDE_CODE_ENABLE_FUNCTION_HOOKS=1. macOS — copy-band shells out to pbcopy, and mini-offload/money-band assume ssh and a Homebrew path on the remote.

License

MIT

Source 1 files
hooks/register.tsx 292 lines
1/* @jsx h */
2import type { EngineInterface, Register } from 'claude-code'
3
4// After each Edit, Write, NotebookEdit or `sed -i` lands, the file's uncommitted hunk sits in a
5// docked pane with [keep] and [revert]: a review the model never sees and a revert that costs no
6// turn. The pane lists the files touched this turn, newest first, and empties at the next turn.
7//
8// A press is answered in `ui.press` by the Button's key, not its closure: a closure belongs to one
9// drawing and a press during a redraw is dropped, while the hook still sees the key.
10
11const MOD = 'diff-review'
12const PANE_ID = 'diff-review'
13const PANE_TITLE = 'diff'
14const ENABLED_KEY = 'diff-review:enabled'
15const AUTO_OPEN_MIN_COLUMNS = 144
16const CODE_MAX_CHARS = 10000
17const GIT_TIMEOUT_MS = 5000
18
19type Hunk = { id: number; path: string; rel: string; diff: string; kind: 'tracked' | 'untracked' | 'new' }
20
21let hunks: Hunk[] = []
22let nextId = 1
23let enabled = true
24let interactive = false
25let cwd = ''
26let columns: number | undefined
27let paneOpen = false
28let paneDrawn = false
29let closedByPerson = false
30let armed: { id: number; timer: { cancel: () => void } } | undefined
31const ARM_MS = 10000
32
33const TOKEN = /'[^']*'|"[^"]*"|\S+/g
34const SED_VALUED = ['-e', '-f', '--expression', '--file']
35
36// The files a `sed -i` writes: every segment of the command that starts with sed and carries -i,
37// its flags dropped, the script dropped unless -e or -f named it.
38export const sedTargets = (command: string): string[] => {
39  const out: string[] = []
40  for (const seg of command.split(/&&|\|\||;|\||\n/)) {
41    const tokens = (seg.match(TOKEN) ?? []).filter(t => t !== "''" && t !== '""')
42    const at = tokens.findIndex(t => t === 'sed')
43    if (at < 0 || !tokens.slice(at + 1).some(t => /^(-i|--in-place)/.test(t))) continue
44    const rest: string[] = []
45    let scripted = false
46    for (let i = at + 1; i < tokens.length; i++) {
47      const t = tokens[i] ?? ''
48      if (SED_VALUED.includes(t)) {
49        scripted = true
50        i++
51      } else if (t.startsWith('-')) continue
52      else rest.push(t.replace(/^(['"])(.*)\1$/, '$2'))
53    }
54    out.push(...(scripted ? rest : rest.slice(1)).filter(t => t !== '' && !/[*?$]/.test(t)))
55  }
56  return out
57}
58
59const absolute = (path: string) => (path.startsWith('/') ? path : `${cwd}/${path}`)
60const relative = (path: string) => (cwd && path.startsWith(`${cwd}/`) ? path.slice(cwd.length + 1) : path)
61
62const addedDiff = (rel: string, text: string) => {
63  const lines = text.replace(/\n$/, '').split('\n')
64  return [`--- /dev/null`, `+++ b/${rel}`, `@@ -0,0 +1,${lines.length} @@`, ...lines.map(l => `+${l}`)].join('\n')
65}
66
67// A Code source holds at most 10000 characters; a hunk past that, or past the rows the pane has,
68// keeps its head and says how much was cut.
69export const fit = (diff: string, maxLines: number, maxChars = CODE_MAX_CHARS): string => {
70  const lines = diff.split('\n')
71  const kept: string[] = []
72  let size = 0
73  for (const line of lines) {
74    const note = `… (${lines.length - kept.length - 1} more lines)`
75    if (kept.length >= maxLines || size + line.length + 1 + note.length + 1 > maxChars) break
76    kept.push(line)
77    size += line.length + 1
78  }
79  const dropped = lines.length - kept.length
80  return dropped > 0 ? [...kept, `… (${dropped} more lines)`].join('\n') : diff
81}
82
83const git = ($: EngineInterface, args: string[]) => $.process.run(['git', ...args], { timeoutMs: GIT_TIMEOUT_MS })
84
85// An untracked file has no index to diff against: one the tool created shows as added and may be
86// deleted; one that was already there (an .env, a gitignored file) shows as added with no revert.
87const hunkOf = async ($: EngineInterface, path: string, existed: boolean): Promise<Hunk | undefined> => {
88  const rel = relative(path)
89  const tracked = await git($, ['ls-files', '--error-unmatch', '--', path])
90  if (tracked.exitCode === 0) {
91    const run = await git($, ['diff', '--unified=3', '--', path])
92    if (run.exitCode !== 0 || !run.stdout.trim()) return undefined
93    return { id: nextId++, path, rel, diff: run.stdout.replace(/\n$/, ''), kind: 'tracked' }
94  }
95  if (tracked.exitCode !== 1) return undefined
96  const text = await $.fs.read(path).catch(() => undefined)
97  if (typeof text !== 'string') return undefined
98  return { id: nextId++, path, rel, diff: addedDiff(rel, text), kind: existed ? 'untracked' : 'new' }
99}
100
101const canOpen = () => enabled && interactive && !paneOpen && !closedByPerson && columns !== undefined && columns >= AUTO_OPEN_MIN_COLUMNS
102
103const open = async ($: EngineInterface) => {
104  paneOpen = true
105  paneDrawn = false
106  await $.ui.open({ id: PANE_ID, title: PANE_TITLE }).catch(err => {
107    paneOpen = false
108    $.ui.log(`diff-review: pane did not open: ${err}`)
109  })
110}
111
112const track = async ($: EngineInterface, paths: string[], existed: boolean[]) => {
113  for (const [i, path] of paths.entries()) {
114    const hunk = await hunkOf($, path, existed[i] === true)
115    hunks = hunks.filter(h => h.path !== path)
116    if (hunk) hunks.unshift(hunk)
117  }
118  if (canOpen()) await open($)
119  $.ui.invalidate('ui.render')
120}
121
122const drop = (id: number) => {
123  hunks = hunks.filter(h => h.id !== id)
124  if (armed?.id === id) disarm()
125}
126
127const disarm = () => {
128  armed?.timer.cancel()
129  armed = undefined
130}
131
132// Only the captured hunk is undone, never the file's other uncommitted changes. A file the tool
133// created is deleted on a second press within ten seconds: no dialog waits inside the hook, so a
134// dropped hook can never leave an rm behind.
135const revert = async ($: EngineInterface, hunk: Hunk) => {
136  if (hunk.kind === 'new') {
137    if (armed?.id !== hunk.id) {
138      disarm()
139      armed = { id: hunk.id, timer: $.clock.after(ARM_MS, () => { armed = undefined; $.ui.invalidate('ui.render') }) }
140      return $.ui.toast(`${hunk.rel} did not exist before this turn: press revert again within 10s to delete it`, { timeoutMs: ARM_MS })
141    }
142    disarm()
143    const rm = await $.process.run(['rm', '--', hunk.path], { timeoutMs: GIT_TIMEOUT_MS })
144    if (rm.exitCode !== 0) return $.ui.toast(`diff-review: delete failed: ${rm.stderr.trim()}`, { timeoutMs: 6000 })
145    drop(hunk.id)
146    return $.ui.toast(`deleted ${hunk.rel}`)
147  }
148  const run = await $.process.run(['git', 'apply', '-R', '--unidiff-zero', '--'], { stdin: `${hunk.diff}\n`, timeoutMs: GIT_TIMEOUT_MS })
149  if (run.exitCode !== 0) return $.ui.toast('hunk no longer applies; nothing changed', { timeoutMs: 6000 })
150  drop(hunk.id)
151  $.ui.toast(`reverted ${hunk.rel}`)
152}
153
154// The first thing anyone does when a mod misbehaves is try to turn it off. `CLAUDE_MODS_DISABLE=all`,
155// or a comma list naming this mod, makes every hook here a pass-through and registers no command.
156let disabled = false
157const readDisabled = async ($: EngineInterface): Promise<boolean> => {
158  const raw = (await $.env.get('CLAUDE_MODS_DISABLE').catch(() => undefined)) ?? ''
159  disabled = raw
160    .split(',')
161    .map(v => v.trim())
162    .some(v => v === 'all' || v === MOD)
163  return disabled
164}
165
166export const register: Register = on => {
167  on('session.start', async ($, e, next) => {
168    const r = await next(e)
169    if (await readDisabled($)) return r
170    interactive = e.isInteractive
171    cwd = e.cwd
172    enabled = (await $.store.get(ENABLED_KEY).catch(() => undefined)) !== false
173    await $.command
174      .register({
175        name: 'diff-review',
176        description: 'The diff pane: each edited file\'s hunk with [keep] and [revert] (diff-review)',
177        argumentHint: '[open | close | on | off | status]',
178        immediate: true,
179      })
180      .catch(err => $.ui.log(`diff-review: /diff-review not registered: ${err}`))
181    return r
182  })
183
184  on('command.run', { command: 'diff-review' }, async ($, e) => {
185    const arg = e.args.trim().toLowerCase()
186    if (arg === 'open') {
187      closedByPerson = false
188      await open($)
189      return { text: paneOpen ? `diff-review pane open, ${hunks.length} file(s) this turn` : 'diff-review: pane did not open' }
190    }
191    if (arg === 'close') {
192      await $.ui.close({ id: PANE_ID }).catch(() => undefined)
193      paneOpen = false
194      return { text: 'diff-review pane closed' }
195    }
196    if (arg === 'on' || arg === 'off') {
197      enabled = arg === 'on'
198      await $.store.set(ENABLED_KEY, enabled).catch(err => $.ui.log(`diff-review: store write failed: ${err}`))
199      if (!enabled && paneOpen) {
200        await $.ui.close({ id: PANE_ID }).catch(() => undefined)
201        paneOpen = false
202      }
203      return { text: `diff-review ${arg}` }
204    }
205    if (arg === '' || arg === 'status') {
206      return { text: `diff-review is ${enabled ? 'on' : 'off'}; pane ${paneOpen ? 'open' : 'closed'}; ${hunks.length} file(s) this turn` }
207    }
208    return { text: `diff-review: "${arg}" is not open, close, on, off, or status` }
209  })
210
211  on('ui.render', { component: 'PromptHint' }, ($, e, next) => {
212    if (e.surface === 'terminal') columns = e.viewport?.columns ?? columns
213    return next(e)
214  })
215
216  // An open the plugin made on its own waits undrawn on a narrow terminal; a prompt is the person's
217  // input, so an open answering it is placed. Not awaited: a refused open cannot delay the turn.
218  on('prompt.submit', ($, e, next) => {
219    if (!disabled && paneOpen && !paneDrawn) void $.ui.open({ id: PANE_ID, title: PANE_TITLE }).catch(() => undefined)
220    return next(e)
221  })
222
223  on('turn.start', ($, e, next) => {
224    if (disabled || !hunks.length) return next(e)
225    hunks = []
226    $.ui.invalidate('ui.render')
227    return next(e)
228  })
229
230  on('ui.close', { id: PANE_ID }, ($, e, next) => {
231    paneOpen = false
232    if (e.origin.kind === 'person') closedByPerson = true
233    return next(e)
234  })
235
236  on('tool.call', { tool: ['Edit', 'Write', 'NotebookEdit', 'Bash'] }, async ($, e, next) => {
237    const off = disabled || !enabled || e.agentId !== undefined
238    const paths = off
239      ? []
240      : e.tool === 'Bash' ? sedTargets(e.command).map(absolute) : e.tool === 'NotebookEdit' ? [e.notebook_path] : [e.file_path]
241    const existed = await Promise.all(paths.map(p => $.fs.exists(p).catch(() => true)))
242    const r = await next(e)
243    if (!paths.length || r.deny !== undefined || r.isError === true) return r
244    try {
245      await track($, paths, existed)
246    } catch (err) {
247      $.ui.log(`diff-review: diff not read: ${err}`)
248    }
249    return r
250  })
251
252  on('ui.press', { plugin: MOD }, async ($, e, next) => {
253    if (disabled) return next(e)
254    const m = /^(keep|revert):(\d+)$/.exec(e.element)
255    const hunk = m ? hunks.find(h => h.id === Number(m[2])) : undefined
256    if (!m || !hunk) return next(e)
257    if (m[1] === 'keep') drop(hunk.id)
258    else await revert($, hunk)
259    $.ui.invalidate('ui.render')
260    return { element: e.element }
261  })
262
263  on('ui.render', { component: 'Pane' }, async ($, e, next) => {
264    if (disabled || e.requestId !== PANE_ID) return next(e)
265    paneDrawn = true
266    const { Box, Text, Button, Code } = await $.ui.resolve(e)
267    const width = Math.max(1, e.props.bodyColumns)
268    const rows = Math.max(3, e.props.scroll.bodyRows - 2)
269    if (!hunks.length) return <Text dimColor>no edits this turn</Text>
270    return (
271      <Box flexDirection="column">
272        {hunks.map(hunk => (
273          <Box key={`hunk:${hunk.id}`} flexDirection="column" marginTop={1}>
274            <Box flexDirection="row" columnGap={1}>
275              <Text bold wrap="truncate-end">
276                {hunk.rel.length > width ? hunk.rel.slice(0, width) : hunk.rel}
277              </Text>
278              <Button key={`keep:${hunk.id}`} label="keep" onPress={() => undefined} />
279              {hunk.kind === 'untracked' ? (
280                <Text dimColor>untracked, existed before: no revert</Text>
281              ) : (
282                <Button key={`revert:${hunk.id}`} label={armed?.id === hunk.id ? 'revert again to delete' : hunk.kind === 'new' ? 'delete' : 'revert'} onPress={() => undefined} />
283              )}
284            </Box>
285            <Code source={fit(hunk.diff, rows)} format="diff" path={hunk.path} />
286          </Box>
287        ))}
288      </Box>
289    )
290  })
291}
292