diff --git a/docs/claude-fable-weekly-usage-meter.md b/docs/claude-fable-weekly-usage-meter.md
index 4a37fee43e9..ae7bcc0f909 100644
--- a/docs/claude-fable-weekly-usage-meter.md
+++ b/docs/claude-fable-weekly-usage-meter.md
@@ -2,7 +2,7 @@
 
 ## Problem
 
-Claude Code now exposes weekly subscription usage alongside the 5-hour window, and its live `/usage` panel can also show a Fable-specific weekly bucket. Anthropic documents `rate_limits.five_hour` and `rate_limits.seven_day` in Claude Code statusline JSON, with weekly data present for Claude.ai subscribers after the first API response. The existing Orca Claude meter already has a weekly slot in shared state, but it needs a distinct Fable weekly slot so the status bar can show all three visible meters when Claude reports them.
+Claude Code now exposes weekly subscription usage alongside the 5-hour window, and its live `/usage` panel can show an explicit Fable-specific weekly bucket. Anthropic documents `rate_limits.five_hour` and `rate_limits.seven_day` in Claude Code statusline JSON, with weekly data present for Claude.ai subscribers after the first API response. The existing Orca Claude meter already has a weekly slot in shared state, but it needs a distinct Fable weekly slot so the status bar can show all three visible meters when Claude reports them.
 
 Relevant code:
 
@@ -17,6 +17,25 @@ Research:
 - Official Claude Code statusline docs: [`rate_limits.five_hour.used_percentage` and `rate_limits.seven_day.used_percentage`](https://code.claude.com/docs/en/statusline#available-data), plus matching `resets_at`, are the 5-hour and 7-day rate-limit fields.
 - `harveyxiacn/cc-usage-monitor` uses Claude Code's statusline `rate_limits` data and shows both [`5h` and `7d` windows](https://github.com/harveyxiacn/cc-usage-monitor), matching Orca's existing `session` and `weekly` model.
 - `leeguooooo/claude-code-usage-bar` independently exposes the same [`5h` and `7d` rate-limit usage](https://github.com/leeguooooo/claude-code-usage-bar) in a Claude Code statusLine integration.
+- Fable is not part of the documented statusline schema above. Orca only treats it as Fable weekly usage when the live `/usage` panel renders a standalone `Fable` label or an OAuth response uses an explicit weekly/seven-day Fable field name.
+
+Redacted live `/usage` shape this parser targets:
+
+```text
+Plan usage limits
+
+Current session
+18% remaining
+Resets in 2h 10m
+
+Current week (all models)
+84% left
+Resets in 5d 4h
+
+Fable
+42% consumed
+Resets in 3d 2h
+```
 
 ## Goal
 
@@ -34,7 +53,7 @@ Make Orca's existing Claude status-bar meter show the weekly Claude and Fable us
 2. Add `ProviderRateLimits.fableWeekly` as a distinct optional Claude window so the chip and popover can render Session, Weekly, and Fable simultaneously.
 3. Accept both OAuth `utilization` windows and Claude Code-style `used_percentage` windows with epoch-second `resets_at` values.
 4. Broaden the hidden Claude CLI parser so the generic weekly label accepts both old `Current week` wording and newer usage/statusline wording: `Weekly limits`, `Weekly usage`, `weekly rate limit`, and `7-day`.
-5. Parse an explicit `Fable` label into `fableWeekly` instead of collapsing it into generic `weekly`.
+5. Parse only a standalone `Fable` label into `fableWeekly` instead of collapsing it into generic `weekly`; ambiguous Fable copy is a section boundary, not a meter.
 6. Broaden percent parsing to treat `consumed` like `used`, because Anthropic describes rate-limit percentages as consumed.
 7. Add focused tests for the new weekly wording and retain existing old-copy coverage.
 
@@ -43,7 +62,8 @@ Make Orca's existing Claude status-bar meter show the weekly Claude and Fable us
 - Weekly data may be absent for API-key users or before the first Claude API response; keep `weekly: null`.
 - The hidden PTY fallback may still only return session data; the status bar should continue showing the 5-hour meter without error.
 - Reset timestamps/descriptions may be absent from CLI output; keep `resetsAt: null` and parse only visible reset text.
-- Fable data may be absent from the documented statusline payload even when the interactive `/usage` panel shows it; keep `fableWeekly: null` unless a distinct field or explicit `Fable` label is present.
+- Fable data may be absent from the documented statusline payload even when the interactive `/usage` panel shows it; keep `fableWeekly: null` unless an explicit weekly/seven-day field or standalone `Fable` label is present.
+- A bare OAuth `fable` field is ambiguous because it does not encode the window length; ignore it until the upstream contract is clearer.
 
 ## Rollout
 
diff --git a/src/main/rate-limits/claude-fetcher.test.ts b/src/main/rate-limits/claude-fetcher.test.ts
index ed5a971f97c..0790374ce0b 100644
--- a/src/main/rate-limits/claude-fetcher.test.ts
+++ b/src/main/rate-limits/claude-fetcher.test.ts
@@ -205,6 +205,42 @@ describe('fetchClaudeRateLimits', () => {
     })
   })
 
+  it('ignores bare Fable OAuth usage because the window length is ambiguous', async () => {
+    const configDir = '/Users/test/.claude'
+    const authPreparation: ClaudeRuntimeAuthPreparation = {
+      configDir,
+      envPatch: { CLAUDE_CONFIG_DIR: configDir },
+      stripAuthEnv: false,
+      provenance: 'system'
+    }
+    vi.mocked(readActiveClaudeKeychainCredentialsStrict).mockResolvedValueOnce(
+      JSON.stringify({
+        claudeAiOauth: {
+          accessToken: 'oauth-token',
+          expiresAt: Date.now() + 60_000
+        }
+      })
+    )
+    netFetchMock.mockResolvedValueOnce(
+      new Response(
+        JSON.stringify({
+          five_hour: { utilization: 11 },
+          seven_day: { utilization: 22 },
+          fable: { utilization: 33 }
+        }),
+        { status: 200 }
+      )
+    )
+
+    await expect(fetchClaudeRateLimits({ authPreparation })).resolves.toMatchObject({
+      provider: 'claude',
+      status: 'ok',
+      session: { usedPercent: 11 },
+      weekly: { usedPercent: 22 },
+      fableWeekly: null
+    })
+  })
+
   it('falls back to legacy Keychain credentials for host system default without an explicit config dir', async () => {
     const configDir = '/Users/test/.claude'
     const authPreparation: ClaudeRuntimeAuthPreparation = {
diff --git a/src/main/rate-limits/claude-fetcher.ts b/src/main/rate-limits/claude-fetcher.ts
index e9c72fbbe11..c89aef0473c 100644
--- a/src/main/rate-limits/claude-fetcher.ts
+++ b/src/main/rate-limits/claude-fetcher.ts
@@ -300,7 +300,6 @@ type OAuthUsageWindow = {
 type OAuthUsageResponse = {
   five_hour?: OAuthUsageWindow
   seven_day?: OAuthUsageWindow
-  fable?: OAuthUsageWindow
   fable_weekly?: OAuthUsageWindow
   fable_seven_day?: OAuthUsageWindow
   seven_day_fable?: OAuthUsageWindow
@@ -378,13 +377,12 @@ function mapWindow(
 }
 
 function mapFableWeeklyWindow(data: OAuthUsageResponse): RateLimitWindow | null {
-  // Why: the Fable weekly bucket is newer than the documented statusline
-  // contract, so accept the observed likely field names without guessing usage.
+  // Why: a bare "fable" field does not prove the window length. Only accept
+  // explicit weekly/seven-day names for the distinct Fable meter.
   return (
     mapWindow(data.fable_weekly, 10080) ??
     mapWindow(data.fable_seven_day, 10080) ??
-    mapWindow(data.seven_day_fable, 10080) ??
-    mapWindow(data.fable, 10080)
+    mapWindow(data.seven_day_fable, 10080)
   )
 }
 
diff --git a/src/main/rate-limits/claude-pty.test.ts b/src/main/rate-limits/claude-pty.test.ts
index 977eec808b5..a613194868e 100644
--- a/src/main/rate-limits/claude-pty.test.ts
+++ b/src/main/rate-limits/claude-pty.test.ts
@@ -258,6 +258,112 @@ describe('fetchViaPty', () => {
     })
   })
 
+  it('does not let an incomplete Fable section consume later usage sections', async () => {
+    const term = makeMockTerm()
+    spawnMock.mockReturnValue(term)
+
+    const resultPromise = fetchViaPty()
+
+    await vi.advanceTimersByTimeAsync(2_000)
+    term.emitData(`
+      Plan usage limits
+
+      Fable
+      Usage unavailable
+
+      Current session
+      12% used
+
+      Current week (all models)
+      84% left
+    `)
+    await vi.advanceTimersByTimeAsync(2_000)
+
+    await expect(resultPromise).resolves.toMatchObject({
+      provider: 'claude',
+      status: 'ok',
+      session: {
+        usedPercent: 12
+      },
+      weekly: {
+        usedPercent: 16
+      },
+      fableWeekly: null,
+      error: null
+    })
+  })
+
+  it('does not treat inline Fable weekly text as a parsed usage label', async () => {
+    const term = makeMockTerm()
+    spawnMock.mockReturnValue(term)
+
+    const resultPromise = fetchViaPty()
+
+    await vi.advanceTimersByTimeAsync(2_000)
+    term.emitData(`
+      Plan usage limits
+
+      Current session
+      Fable weekly usage
+      42% consumed
+
+      Current week (all models)
+      84% left
+    `)
+    await vi.advanceTimersByTimeAsync(2_000)
+
+    await expect(resultPromise).resolves.toMatchObject({
+      provider: 'claude',
+      status: 'ok',
+      session: null,
+      weekly: {
+        usedPercent: 16
+      },
+      fableWeekly: null,
+      error: null
+    })
+  })
+
+  it('keeps waiting after a bare Fable heading until another usage section renders', async () => {
+    const term = makeMockTerm()
+    spawnMock.mockReturnValue(term)
+    let settled = false
+
+    const resultPromise = fetchViaPty().finally(() => {
+      settled = true
+    })
+
+    await vi.advanceTimersByTimeAsync(2_000)
+    term.emitData(`
+      Plan usage limits
+
+      Fable
+    `)
+    await vi.advanceTimersByTimeAsync(2_000)
+
+    expect(settled).toBe(false)
+
+    term.emitData(`
+      42% consumed
+
+      Current session
+      12% used
+    `)
+    await vi.advanceTimersByTimeAsync(2_000)
+
+    await expect(resultPromise).resolves.toMatchObject({
+      provider: 'claude',
+      status: 'ok',
+      session: {
+        usedPercent: 12
+      },
+      fableWeekly: {
+        usedPercent: 42
+      },
+      error: null
+    })
+  })
+
   it('parses 7-day weekly labels without the old Current week heading', async () => {
     const term = makeMockTerm()
     spawnMock.mockReturnValue(term)
diff --git a/src/main/rate-limits/claude-pty.ts b/src/main/rate-limits/claude-pty.ts
index b2e2caa3a3b..f0363eba3f2 100644
--- a/src/main/rate-limits/claude-pty.ts
+++ b/src/main/rate-limits/claude-pty.ts
@@ -20,7 +20,8 @@ const MAX_OUTPUT_LENGTH = 100_000 // 100KB buffer limit
 // Weekly labels have varied between "Current week" and "Weekly limits".
 const SESSION_RE = /current\s*session/i
 const WEEKLY_RE = /(?:current\s*week|weekly\s*(?:limits?|usage|rate\s*limits?)|7\s*[- ]?\s*day)/i
-const FABLE_RE = /\bfable\b/i
+const FABLE_WORD_RE = /\bfable\b/i
+const FABLE_LABEL_RE = /^\s*fable\s*$/i
 const PERCENT_RE = /(\d{1,3})(?:\.\d+)?\s*%\s*(used|consumed|left|remaining|available)/i
 const RESET_LINE_RE = /resets?\s+(?:at\s+|in\s+)?(.+)/i
 const ESC = String.fromCharCode(27)
@@ -36,18 +37,29 @@ function stripTerminalControlSequences(output: string): string {
  * Extract percent-left from lines following a label match.
  * Scans up to 12 lines after the label to find the associated percent.
  */
+function matchesWeeklyLabel(line: string): boolean {
+  return WEEKLY_RE.test(line) && !FABLE_WORD_RE.test(line)
+}
+
+function matchesFableBoundary(line: string): boolean {
+  return FABLE_LABEL_RE.test(line) || (FABLE_WORD_RE.test(line) && WEEKLY_RE.test(line))
+}
+
+function isSectionLabel(line: string): boolean {
+  return SESSION_RE.test(line) || matchesWeeklyLabel(line) || matchesFableBoundary(line)
+}
+
 function extractPercentAfterLabel(
   lines: string[],
-  labelRe: RegExp,
-  options: { stopAtFable?: boolean } = {}
+  matchesLabel: (line: string) => boolean
 ): number | null {
   for (let i = 0; i < lines.length; i++) {
-    if (!labelRe.test(lines[i])) {
+    if (!matchesLabel(lines[i])) {
       continue
     }
     // Scan next 12 lines for a percent
     for (let j = i; j < Math.min(i + 12, lines.length); j++) {
-      if (j > i && options.stopAtFable === true && FABLE_RE.test(lines[j])) {
+      if (j > i && isSectionLabel(lines[j])) {
         break
       }
       const m = PERCENT_RE.exec(lines[j])
@@ -64,15 +76,14 @@ function extractPercentAfterLabel(
 
 function extractResetAfterLabel(
   lines: string[],
-  labelRe: RegExp,
-  options: { stopAtFable?: boolean } = {}
+  matchesLabel: (line: string) => boolean
 ): string | null {
   for (let i = 0; i < lines.length; i++) {
-    if (!labelRe.test(lines[i])) {
+    if (!matchesLabel(lines[i])) {
       continue
     }
     for (let j = i; j < Math.min(i + 14, lines.length); j++) {
-      if (j > i && options.stopAtFable === true && FABLE_RE.test(lines[j])) {
+      if (j > i && isSectionLabel(lines[j])) {
         break
       }
       const m = RESET_LINE_RE.exec(lines[j])
@@ -91,9 +102,9 @@ function parsePtyUsage(output: string): {
 } {
   const lines = output.split(/\r\n|\n|\r/)
 
-  const sessionPct = extractPercentAfterLabel(lines, SESSION_RE)
-  const weeklyPct = extractPercentAfterLabel(lines, WEEKLY_RE, { stopAtFable: true })
-  const fableWeeklyPct = extractPercentAfterLabel(lines, FABLE_RE)
+  const sessionPct = extractPercentAfterLabel(lines, (line) => SESSION_RE.test(line))
+  const weeklyPct = extractPercentAfterLabel(lines, matchesWeeklyLabel)
+  const fableWeeklyPct = extractPercentAfterLabel(lines, (line) => FABLE_LABEL_RE.test(line))
 
   const session: RateLimitWindow | null =
     sessionPct !== null
@@ -101,7 +112,7 @@ function parsePtyUsage(output: string): {
           usedPercent: Math.min(100, Math.max(0, sessionPct)),
           windowMinutes: 300,
           resetsAt: null,
-          resetDescription: extractResetAfterLabel(lines, SESSION_RE)
+          resetDescription: extractResetAfterLabel(lines, (line) => SESSION_RE.test(line))
         }
       : null
 
@@ -111,7 +122,7 @@ function parsePtyUsage(output: string): {
           usedPercent: Math.min(100, Math.max(0, weeklyPct)),
           windowMinutes: 10080,
           resetsAt: null,
-          resetDescription: extractResetAfterLabel(lines, WEEKLY_RE, { stopAtFable: true })
+          resetDescription: extractResetAfterLabel(lines, matchesWeeklyLabel)
         }
       : null
 
@@ -121,7 +132,7 @@ function parsePtyUsage(output: string): {
           usedPercent: Math.min(100, Math.max(0, fableWeeklyPct)),
           windowMinutes: 10080,
           resetsAt: null,
-          resetDescription: extractResetAfterLabel(lines, FABLE_RE)
+          resetDescription: extractResetAfterLabel(lines, (line) => FABLE_LABEL_RE.test(line))
         }
       : null
 
@@ -140,7 +151,6 @@ const STOP_SUBSTRINGS = [
   'Weekly limit',
   'Weekly usage',
   '7-day',
-  'Fable',
   'Current session',
   'Failed to load usage data',
   'failed to load usage data'