Add Fable 5 weekly usage meter for Claude Code

[fredshino]

Summary

Adds a Fable 5 weekly usage meter for Claude Code, matching the new Fable-specific meter that the Claude app now shows.

The /api/oauth/usage endpoint now returns a Fable-scoped weekly limit inside its limits array as a weekly_scoped entry whose scope.model.display_name is "Fable". Previously the poller only read five_hour and seven_day, so this data was ignored.

What it does

• Parsing (poller.rs): parse the limits array and extract the Fable-scoped weekly limit into a new optional UsageData.fable section. Falls back gracefully — if the account has no Fable limit, nothing changes.
• Rendering (window.rs): render the Fable meter as an extra bar in the 7d/weekly row, in a distinct violet accent so it reads separately from the weekly-all bar. Because the widget is height-locked to the taskbar (2 rows), the bar extends horizontally and the widget auto-widens to fit it. A "Fable" heading is drawn in the 5h row directly above the bar so it's clearly labelled.
• Toggle: a checkable Fable entry under the right-click Models submenu (shown only when Claude Code is enabled), persisted via a new show_fable setting (default on). It isn't bound by the "at least one model enabled" constraint since it's an optional extra bar.
• Tray tooltip: the Claude tooltip gains a | Fable 7d: … line when the meter is active.

The meter appears automatically only when the account actually has a Fable-scoped limit.

Notes

• The "Fable" label is hardcoded (a model/brand name) rather than added to the localization Strings, to keep the diff focused — happy to move it into the localization files if you'd prefer.
• Added unit tests for the new parsing (extracts_fable_weekly_scoped_limit, no_fable_section_when_absent).

Testing

• cargo test → all tests pass (incl. the two new ones).
• cargo build --release → clean, no warnings.
• Verified live on Windows 11: the violet Fable bar renders in the 7d row with the "Fable" heading above it, the Models → Fable toggle shows/hides it and persists across restarts.

Screenshot

[fredshino]

Summary

Adds a Fable 5 weekly usage meter for Claude Code, matching the new Fable-specific meter that the Claude app now shows.

The /api/oauth/usage endpoint now returns a Fable-scoped weekly limit inside its limits array as a weekly_scoped entry whose scope.model.display_name is "Fable". Previously the poller only read five_hour and seven_day, so this data was ignored.

What it does

• Parsing (poller.rs): parse the limits array and extract the Fable-scoped weekly limit into a new optional UsageData.fable section. Falls back gracefully — if the account has no Fable limit, nothing changes.
• Rendering (window.rs): render the Fable meter as an extra bar in the 7d/weekly row, in a distinct violet accent so it reads separately from the weekly-all bar. Because the widget is height-locked to the taskbar (2 rows), the bar extends horizontally and the widget auto-widens to fit it. A "Fable" heading is drawn in the 5h row directly above the bar so it's clearly labelled.
• Toggle: a checkable Fable entry under the right-click Models submenu (shown only when Claude Code is enabled), persisted via a new show_fable setting (default on). It isn't bound by the "at least one model enabled" constraint since it's an optional extra bar.
• Tray tooltip: the Claude tooltip gains a | Fable 7d: … line when the meter is active.

The meter appears automatically only when the account actually has a Fable-scoped limit.

Notes

• The "Fable" label is hardcoded (a model/brand name) rather than added to the localization Strings, to keep the diff focused — happy to move it into the localization files if you'd prefer.
• Added unit tests for the new parsing (extracts_fable_weekly_scoped_limit, no_fable_section_when_absent).

Testing

• cargo test → all tests pass (incl. the two new ones).
• cargo build --release → clean, no warnings.
• Verified live on Windows 11: the violet Fable bar renders in the 7d row with the "Fable" heading above it, the Models → Fable toggle shows/hides it and persists across restarts.

Screenshot

[fredshino]

Summary

Adds a Fable 5 weekly usage meter for Claude Code, matching the new Fable-specific meter that the Claude app now shows.

The /api/oauth/usage endpoint now returns a Fable-scoped weekly limit inside its limits array as a weekly_scoped entry whose scope.model.display_name is "Fable". Previously the poller only read five_hour and seven_day, so this data was ignored.

What it does

• Parsing (poller.rs): parse the limits array and extract the Fable-scoped weekly limit into a new optional UsageData.fable section. Falls back gracefully — if the account has no Fable limit, nothing changes.
• Rendering (window.rs): render the Fable meter as an extra bar in the 7d/weekly row, in a distinct violet accent so it reads separately from the weekly-all bar. Because the widget is height-locked to the taskbar (2 rows), the bar extends horizontally and the widget auto-widens to fit it. A "Fable" heading is drawn in the 5h row directly above the bar so it's clearly labelled.
• Toggle: a checkable Fable entry under the right-click Models submenu (shown only when Claude Code is enabled), persisted via a new show_fable setting (default on). It isn't bound by the "at least one model enabled" constraint since it's an optional extra bar.
• Tray tooltip: the Claude tooltip gains a | Fable 7d: … line when the meter is active.

The meter appears automatically only when the account actually has a Fable-scoped limit.

Notes

• The "Fable" label is hardcoded (a model/brand name) rather than added to the localization Strings, to keep the diff focused — happy to move it into the localization files if you'd prefer.
• Added unit tests for the new parsing (extracts_fable_weekly_scoped_limit, no_fable_section_when_absent).

Testing

• cargo test → all tests pass (incl. the two new ones).
• cargo build --release → clean, no warnings.
• Verified live on Windows 11: the violet Fable bar renders in the 7d row with the "Fable" heading above it, the Models → Fable toggle shows/hides it and persists across restarts.

Screenshot

[fredshino]

Summary

Adds a Fable 5 weekly usage meter for Claude Code, matching the new Fable-specific meter that the Claude app now shows.

The /api/oauth/usage endpoint now returns a Fable-scoped weekly limit inside its limits array as a weekly_scoped entry whose scope.model.display_name is "Fable". Previously the poller only read five_hour and seven_day, so this data was ignored.

What it does

• Parsing (poller.rs): parse the limits array and extract the Fable-scoped weekly limit into a new optional UsageData.fable section. Falls back gracefully — if the account has no Fable limit, nothing changes.
• Rendering (window.rs): render the Fable meter as an extra bar in the 7d/weekly row, in a distinct violet accent so it reads separately from the weekly-all bar. Because the widget is height-locked to the taskbar (2 rows), the bar extends horizontally and the widget auto-widens to fit it. A "Fable" heading is drawn in the 5h row directly above the bar so it's clearly labelled.
• Toggle: a checkable Fable entry under the right-click Models submenu (shown only when Claude Code is enabled), persisted via a new show_fable setting (default on). It isn't bound by the "at least one model enabled" constraint since it's an optional extra bar.
• Tray tooltip: the Claude tooltip gains a | Fable 7d: … line when the meter is active.

The meter appears automatically only when the account actually has a Fable-scoped limit.

Notes

• The "Fable" label is hardcoded (a model/brand name) rather than added to the localization Strings, to keep the diff focused — happy to move it into the localization files if you'd prefer.
• Added unit tests for the new parsing (extracts_fable_weekly_scoped_limit, no_fable_section_when_absent).

Testing

• cargo test → all tests pass (incl. the two new ones).
• cargo build --release → clean, no warnings.
• Verified live on Windows 11: the violet Fable bar renders in the 7d row with the "Fable" heading above it, the Models → Fable toggle shows/hides it and persists across restarts.

Screenshot

[fredshino]

Summary

Adds a Fable 5 weekly usage meter for Claude Code, matching the new Fable-specific meter that the Claude app now shows.

The /api/oauth/usage endpoint now returns a Fable-scoped weekly limit inside its limits array as a weekly_scoped entry whose scope.model.display_name is "Fable". Previously the poller only read five_hour and seven_day, so this data was ignored.

What it does

• Parsing (poller.rs): parse the limits array and extract the Fable-scoped weekly limit into a new optional UsageData.fable section. Falls back gracefully — if the account has no Fable limit, nothing changes.
• Rendering (window.rs): render the Fable meter as an extra bar in the 7d/weekly row, in a distinct violet accent so it reads separately from the weekly-all bar. Because the widget is height-locked to the taskbar (2 rows), the bar extends horizontally and the widget auto-widens to fit it. A "Fable" heading is drawn in the 5h row directly above the bar so it's clearly labelled.
• Toggle: a checkable Fable entry under the right-click Models submenu (shown only when Claude Code is enabled), persisted via a new show_fable setting (default on). It isn't bound by the "at least one model enabled" constraint since it's an optional extra bar.
• Tray tooltip: the Claude tooltip gains a | Fable 7d: … line when the meter is active.

The meter appears automatically only when the account actually has a Fable-scoped limit.

Notes

• The "Fable" label is hardcoded (a model/brand name) rather than added to the localization Strings, to keep the diff focused — happy to move it into the localization files if you'd prefer.
• Added unit tests for the new parsing (extracts_fable_weekly_scoped_limit, no_fable_section_when_absent).

Testing

• cargo test → all tests pass (incl. the two new ones).
• cargo build --release → clean, no warnings.
• Verified live on Windows 11: the violet Fable bar renders in the 7d row with the "Fable" heading above it, the Models → Fable toggle shows/hides it and persists across restarts.

Screenshot

[fredshino]

Summary

Adds a Fable 5 weekly usage meter for Claude Code, matching the new Fable-specific meter that the Claude app now shows.

The /api/oauth/usage endpoint now returns a Fable-scoped weekly limit inside its limits array as a weekly_scoped entry whose scope.model.display_name is "Fable". Previously the poller only read five_hour and seven_day, so this data was ignored.

What it does

• Parsing (poller.rs): parse the limits array and extract the Fable-scoped weekly limit into a new optional UsageData.fable section. Falls back gracefully — if the account has no Fable limit, nothing changes.
• Rendering (window.rs): render the Fable meter as an extra bar in the 7d/weekly row, in a distinct violet accent so it reads separately from the weekly-all bar. Because the widget is height-locked to the taskbar (2 rows), the bar extends horizontally and the widget auto-widens to fit it. A "Fable" heading is drawn in the 5h row directly above the bar so it's clearly labelled.
• Toggle: a checkable Fable entry under the right-click Models submenu (shown only when Claude Code is enabled), persisted via a new show_fable setting (default on). It isn't bound by the "at least one model enabled" constraint since it's an optional extra bar.
• Tray tooltip: the Claude tooltip gains a | Fable 7d: … line when the meter is active.

The meter appears automatically only when the account actually has a Fable-scoped limit.

Notes

• The "Fable" label is hardcoded (a model/brand name) rather than added to the localization Strings, to keep the diff focused — happy to move it into the localization files if you'd prefer.
• Added unit tests for the new parsing (extracts_fable_weekly_scoped_limit, no_fable_section_when_absent).

Testing

• cargo test → all tests pass (incl. the two new ones).
• cargo build --release → clean, no warnings.
• Verified live on Windows 11: the violet Fable bar renders in the 7d row with the "Fable" heading above it, the Models → Fable toggle shows/hides it and persists across restarts.

Screenshot

[fredshino]

Summary

Adds a Fable 5 weekly usage meter for Claude Code, matching the new Fable-specific meter that the Claude app now shows.

The /api/oauth/usage endpoint now returns a Fable-scoped weekly limit inside its limits array as a weekly_scoped entry whose scope.model.display_name is "Fable". Previously the poller only read five_hour and seven_day, so this data was ignored.

What it does

• Parsing (poller.rs): parse the limits array and extract the Fable-scoped weekly limit into a new optional UsageData.fable section. Falls back gracefully — if the account has no Fable limit, nothing changes.
• Rendering (window.rs): render the Fable meter as an extra bar in the 7d/weekly row, in a distinct violet accent so it reads separately from the weekly-all bar. Because the widget is height-locked to the taskbar (2 rows), the bar extends horizontally and the widget auto-widens to fit it. A "Fable" heading is drawn in the 5h row directly above the bar so it's clearly labelled.
• Toggle: a checkable Fable entry under the right-click Models submenu (shown only when Claude Code is enabled), persisted via a new show_fable setting (default on). It isn't bound by the "at least one model enabled" constraint since it's an optional extra bar.
• Tray tooltip: the Claude tooltip gains a | Fable 7d: … line when the meter is active.

The meter appears automatically only when the account actually has a Fable-scoped limit.

Notes

• The "Fable" label is hardcoded (a model/brand name) rather than added to the localization Strings, to keep the diff focused — happy to move it into the localization files if you'd prefer.
• Added unit tests for the new parsing (extracts_fable_weekly_scoped_limit, no_fable_section_when_absent).

Testing

• cargo test → all tests pass (incl. the two new ones).
• cargo build --release → clean, no warnings.
• Verified live on Windows 11: the violet Fable bar renders in the 7d row with the "Fable" heading above it, the Models → Fable toggle shows/hides it and persists across restarts.

Screenshot

[fredshino]

Summary

Adds a Fable 5 weekly usage meter for Claude Code, matching the new Fable-specific meter that the Claude app now shows.

The /api/oauth/usage endpoint now returns a Fable-scoped weekly limit inside its limits array as a weekly_scoped entry whose scope.model.display_name is "Fable". Previously the poller only read five_hour and seven_day, so this data was ignored.

What it does

• Parsing (poller.rs): parse the limits array and extract the Fable-scoped weekly limit into a new optional UsageData.fable section. Falls back gracefully — if the account has no Fable limit, nothing changes.
• Rendering (window.rs): render the Fable meter as an extra bar in the 7d/weekly row, in a distinct violet accent so it reads separately from the weekly-all bar. Because the widget is height-locked to the taskbar (2 rows), the bar extends horizontally and the widget auto-widens to fit it. A "Fable" heading is drawn in the 5h row directly above the bar so it's clearly labelled.
• Toggle: a checkable Fable entry under the right-click Models submenu (shown only when Claude Code is enabled), persisted via a new show_fable setting (default on). It isn't bound by the "at least one model enabled" constraint since it's an optional extra bar.
• Tray tooltip: the Claude tooltip gains a | Fable 7d: … line when the meter is active.

The meter appears automatically only when the account actually has a Fable-scoped limit.

Notes

• The "Fable" label is hardcoded (a model/brand name) rather than added to the localization Strings, to keep the diff focused — happy to move it into the localization files if you'd prefer.
• Added unit tests for the new parsing (extracts_fable_weekly_scoped_limit, no_fable_section_when_absent).

Testing

• cargo test → all tests pass (incl. the two new ones).
• cargo build --release → clean, no warnings.
• Verified live on Windows 11: the violet Fable bar renders in the 7d row with the "Fable" heading above it, the Models → Fable toggle shows/hides it and persists across restarts.

Screenshot