fix(naming): keep the built-in branch-name prompt general

The shipped auto-rename prompt no longer hard-codes style rules (word
count, kebab-case, no prefixes). Users can already override naming via
Source Control AI instructions and the branch-name command template; a
prescriptive default fought those overrides. Git-safe sanitization still
runs after generation.

src/main/agent-hooks/first-work-branch-rename.test.ts

@@ -140,7 +140,7 @@ describe('maybeAutoRenameBranchOnFirstWork', () => {
   })
 
   it('strips a prefix the model leaked into the slug from both branch and display name', async () => {
-    // Model ignored "no prefixes" and echoed `you/worktree-spinner`, which the
+    // Model echoed `you/worktree-spinner`, which the
     // sanitizer folds to `you-worktree-spinner`; without stripping it would
     // double-prefix the branch (`you/you-...`) and show "You worktree spinner".
     generateBranchNameMock.mockResolvedValue({ success: true, slug: 'you-worktree-spinner' })

src/main/agent-hooks/first-work-branch-rename.ts

@@ -283,9 +283,8 @@ async function runAutoRename(
   const username = provider
     ? (await getSshGitUsername(provider, repo.path)) || null
     : (await resolveLocalGitUsername(repo.path)) || null
-  // The model is told not to add a prefix, but sometimes echoes the configured
-  // one (e.g. `tmchow/...`); strip it so it doesn't double-prefix the branch or
-  // leak into the display name.
+  // The model sometimes echoes the configured prefix (e.g. `tmchow/...`); strip
+  // it so it doesn't double-prefix the branch or leak into the display name.
   const slug = stripConfiguredBranchPrefix(
     generated.slug,
     getConfiguredBranchPrefix(settings, username)

src/main/text-generation/commit-message-text-generation.test.ts

@@ -1879,8 +1879,9 @@ describe('generateBranchNameFromContext', () => {
       }
     )
 
-    expect(prompt).toContain('Additional user prompt:')
+    expect(prompt.startsWith('Prefer auth terminology.')).toBe(true)
     expect(prompt).toContain('Prefer auth terminology.')
+    expect(prompt).not.toContain('Additional user prompt:')
   })
 })
 

src/shared/branch-name-from-work.test.ts

@@ -100,6 +100,16 @@ describe('buildBranchNamePrompt', () => {
     expect(prompt).not.toContain("Agent's initial response")
   })
 
+  it('keeps the default prompt general without style rules', () => {
+    const prompt = buildBranchNamePrompt({ firstPrompt: 'Add a logout button' })
+    expect(prompt).toContain('Generate a short git branch name')
+    expect(prompt).toContain('Output ONLY the branch name on a single line')
+    expect(prompt).not.toContain('Rules:')
+    expect(prompt).not.toMatch(/kebab-case/i)
+    expect(prompt).not.toMatch(/between \d+ and \d+ words/i)
+    expect(prompt).not.toMatch(/no prefixes/i)
+  })
+
   it('includes the assistant response when present', () => {
     const prompt = buildBranchNamePrompt({
       firstPrompt: 'Add a logout button',
@@ -109,12 +119,14 @@ describe('buildBranchNamePrompt', () => {
     expect(prompt).toContain("I'll wire it into the header.")
   })
 
-  it('appends a custom branch-name prompt when present', () => {
+  it('leads with a custom naming prompt so overrides can own style', () => {
     const prompt = buildBranchNamePrompt(
       { firstPrompt: 'Add a logout button' },
       'Prefer product nouns.'
     )
-    expect(prompt).toContain('Additional user prompt:')
-    expect(prompt).toContain('Prefer product nouns.')
+    expect(prompt.startsWith('Prefer product nouns.')).toBe(true)
+    expect(prompt).not.toContain('Additional user prompt:')
+    expect(prompt).not.toContain('Rules:')
+    expect(prompt).toContain('Add a logout button')
   })
 })

src/shared/branch-name-from-work.ts

@@ -1,10 +1,9 @@
 import { MARINE_CREATURES } from './marine-creatures'
 
-// Why: the work-derived branch name stays short on purpose — long, sentence-like
-// branches read worse than the creature name they replace. Two-to-four words is
-// the sweet spot the feature targets.
+// Why: post-generation sanitization still bounds the leaf so a long model dump
+// cannot become an unreadable branch. The *prompt* stays general — users can
+// override naming style via Source Control AI instructions / templates.
 export const MAX_BRANCH_NAME_WORDS = 4
-const MIN_BRANCH_NAME_WORDS = 2
 
 const CREATURE_NAMES_LOWER = new Set(MARINE_CREATURES.map((name) => name.toLowerCase()))
 
@@ -46,16 +45,15 @@ export function sanitizeBranchSlug(raw: string, maxWords = MAX_BRANCH_NAME_WORDS
 }
 
 /**
- * Drop a leading prefix segment the generation model prepended despite the
- * "no prefixes" instruction — e.g. with a `tmchow/` branch prefix the model
- * emits `tmchow/worktree-spinner`, which `sanitizeBranchSlug` folds to
- * `tmchow-worktree-spinner`. Left alone, that leaks the prefix into both the
- * branch leaf (yielding a doubled `tmchow/tmchow-...`) and the humanized
- * display name. Strips only when the leading segment matches the *configured*
- * prefix, so a work-derived name that merely starts with a real word survives.
- * Prefix-only output (the model echoed just `tmchow`) yields an empty slug so
- * the caller skips the rename — otherwise it would double-prefix to
- * `tmchow/tmchow`.
+ * Drop a leading prefix segment the generation model sometimes prepends (e.g.
+ * with a `tmchow/` branch prefix it emits `tmchow/worktree-spinner`, which
+ * `sanitizeBranchSlug` folds to `tmchow-worktree-spinner`). Left alone, that
+ * leaks the prefix into both the branch leaf (yielding a doubled
+ * `tmchow/tmchow-...`) and the humanized display name. Strips only when the
+ * leading segment matches the *configured* prefix, so a work-derived name that
+ * merely starts with a real word survives. Prefix-only output (the model
+ * echoed just `tmchow`) yields an empty slug so the caller skips the rename —
+ * otherwise it would double-prefix to `tmchow/tmchow`.
  */
 export function stripConfiguredBranchPrefix(
   slug: string,
@@ -100,30 +98,32 @@ export type BranchNameWorkContext = {
 }
 
 /**
- * Build the text-generation prompt that asks the configured agent to summarize
- * the work into a branch name. Kept in shared so the prompt is identical across
- * local and SSH generation targets.
+ * Build the text-generation prompt that asks the configured agent to name the
+ * work. Kept in shared so the prompt is identical across local and SSH targets.
+ *
+ * Why: the shipped default stays general. Naming style is left open so users
+ * who override via Source Control AI instructions / the branch-name command
+ * template are not fighting a rigid built-in rule list. Git-safety (kebab,
+ * length) is enforced after generation by sanitizeBranchSlug.
  */
 export function buildBranchNamePrompt(context: BranchNameWorkContext, customPrompt = ''): string {
-  const sections = [
-    'Generate a git branch name that summarizes the coding task described below.',
-    'Rules:',
-    `- Use between ${MIN_BRANCH_NAME_WORDS} and ${MAX_BRANCH_NAME_WORDS} words.`,
-    '- Lowercase kebab-case only (words joined by single hyphens).',
-    '- No slashes, no prefixes, no quotes, no trailing punctuation.',
-    '- Describe the work itself, not the agent or the repository.',
-    '- Output ONLY the branch name on a single line, nothing else.',
-    '',
-    'User request:',
-    context.firstPrompt.trim()
-  ]
+  const sections: string[] = []
+  const prompt = customPrompt.trim()
+  // Why: when the user supplies naming guidance, lead with it so their
+  // override owns style rather than sitting under a prescriptive default.
+  if (prompt) {
+    sections.push(prompt, '')
+  } else {
+    sections.push(
+      'Generate a short git branch name that summarizes the coding task described below.',
+      'Output ONLY the branch name on a single line, nothing else.',
+      ''
+    )
+  }
+  sections.push('User request:', context.firstPrompt.trim())
   const assistant = context.assistantMessage?.trim()
   if (assistant) {
     sections.push('', "Agent's initial response:", assistant)
   }
-  const prompt = customPrompt.trim()
-  if (prompt) {
-    sections.push('', 'Additional user prompt:', prompt)
-  }
   return sections.join('\n')
 }