Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Remove a feature flag after it has been rolled out and stabilized in the Warp codebase.
.claude/skills/warpdotdev-remove-feature-flag/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 29% | 0% |
Remove a feature flag after it has been rolled out and stabilized in the Warp codebase.
After a feature flag has been enabled for all users and has stabilized in production, the flag should be removed to reduce technical debt and simplify the codebase. This involves removing the flag definition and all conditional checks.
Remove a feature flag when:
default features in app/Cargo.tomlRemove the feature from both the [features] section and the default array:
toml[features] default = [ # Remove "your_feature_name" from here ] # Remove this line: # your_feature_name = []
Remove the variant from the FeatureFlag enum in warp_core/src/features.rs:
rust#[derive(Sequence)] pub enum FeatureFlag { // Remove YourFeatureName, }
Remove the conditional compilation directive:
rust// Remove these lines: // #[cfg(feature = "your_feature_name")] // YourFeatureName,
If the flag was listed in any of these arrays in features.rs, remove it:
rustpub const DOGFOOD_FLAGS: &[FeatureFlag] = &[ // Remove FeatureFlag::YourFeatureName, ];
Find and remove all FeatureFlag::YourFeatureName.is_enabled() checks throughout the codebase:
Before:
rustif FeatureFlag::YourFeatureName.is_enabled() { // new behavior } else { // old behavior (dead code) }
After:
rust// new behavior (unconditionally enabled)
Use ripgrep to find all occurrences. The shared FeatureFlag may be used from the headless TUI, so search crates/warp_tui/ (and other non-app/ crates), not just app/ and warp_core/:
bashrg "YourFeatureName" app/ warp_core/ crates/warp_tui/
If the feature flag was used in keybinding enabled predicates, remove the predicate:
Before:
rustEditableBinding::new( "action:name", "Action description", YourAction::Variant ) .with_enabled(|| FeatureFlag::YourFeatureName.is_enabled()) .with_key_binding("cmdorctrl-key")
After:
rustEditableBinding::new( "action:name", "Action description", YourAction::Variant ) .with_key_binding("cmdorctrl-key")
Remove any code paths that were only executed when the feature was disabled (the else branches in feature checks). These are now dead code.
After removing the flag:
bash# Run affected tests first cargo nextest run -p <affected-package> # Then run the applicable Clippy check cargo clippy -p <affected-package> --all-targets --tests -- -D warnings # Format once after the code is settled ./script/format
Add affected packages or test filters when the flag crosses package boundaries.
Do not run the full workspace suite, launch the GUI or TUI, rerun earlier checks after formatting, or add ./script/presubmit unless the user, task, or approved spec explicitly requires it.
CI owns broader platform and workspace coverage.
bash# Find all occurrences of the flag name (include the TUI and other non-app crates) rg "YourFeatureName" app/ warp_core/ crates/warp_tui/ # Find feature flag checks rg "FeatureFlag::YourFeatureName" app/ crates/warp_tui/ # Find cfg attributes rg 'cfg\(feature = "your_feature_name"\)' app/
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 7,850 | 6,574 | -16% | 1 | 1 | 0% | 211 | 1,490 | +606% | 0 | 0 | — |
case-02 | fail→fail | 5,089 | 7,273 | +43% | 1 | 1 | 0% | 209 | 1,597 | +664% | 0 | 0 | — |
case-03 | fail→fail | 5,844 | 8,201 | +40% | 1 | 1 | 0% | 234 | 1,552 | +563% | 0 | 0 | — |
case-04 | fail→pass | 14,223 | 8,915 | -37% | 1 | 1 | 0% | 2,452 | 2,575 | +5% | 0 | 0 | — |
case-05 | pass→pass | 14,100 | 14,330 | +2% | 1 | 1 | 0% | 1,930 | 2,956 | +53% | 0 | 0 | — |
case-06 | pass→pass | 8,263 | 6,589 | -20% | 1 | 1 | 0% | 1,226 | 2,056 | +68% | 0 | 0 | — |
case-07 | fail→pass | 14,400 | 8,418 | -42% | 1 | 1 | 0% | 2,200 | 2,263 | +3% | 0 | 0 | — |
case-08 | fail→pass | 10,484 | 5,907 | -44% | 1 | 1 | 0% | 1,609 | 1,794 | +11% | 0 | 0 | — |
case-09 | fail→pass | 10,381 | 3,681 | -65% | 1 | 1 | 0% | 1,381 | 1,651 | +20% | 0 | 0 | — |
case-10 | fail→pass | 9,412 | 6,647 | -29% | 1 | 1 | 0% | 1,481 | 1,917 | +29% | 0 | 0 | — |
case-11 | fail→pass | 10,098 | 5,076 | -50% | 1 | 1 | 0% | 1,513 | 1,919 | +27% | 0 | 0 | — |
case-12 | fail→pass | 6,889 | 3,585 | -48% | 1 | 1 | 0% | 1,026 | 1,576 | +54% | 0 | 0 | — |
case-13 | fail→pass | 9,681 | 7,216 | -25% | 1 | 1 | 0% | 1,594 | 1,937 | +22% | 0 | 0 | — |
case-14 | pass→pass | 11,517 | 5,659 | -51% | 1 | 1 | 0% | 1,871 | 2,050 | +10% | 0 | 0 | — |
case-15 | pass→pass | 12,106 | 4,260 | -65% | 1 | 1 | 0% | 1,447 | 1,687 | +17% | 0 | 0 | — |
case-16 | pass→pass | 8,802 | 6,020 | -32% | 1 | 1 | 0% | 1,461 | 1,791 | +23% | 0 | 0 | — |
case-17 | fail→pass | 10,324 | 3,521 | -66% | 1 | 1 | 0% | 1,555 | 1,565 | +1% | 0 | 0 | — |
case-18 | fail→pass | 7,699 | 4,372 | -43% | 1 | 1 | 0% | 1,078 | 1,687 | +56% | 0 | 0 | — |
case-19 | pass→pass | 9,827 | 2,889 | -71% | 1 | 1 | 0% | 1,587 | 1,418 | -11% | 0 | 0 | — |
case-20 | fail→pass | 11,589 | 4,351 | -62% | 1 | 1 | 0% | 1,558 | 1,607 | +3% | 0 | 0 | — |
case-21 | pass→pass | 8,265 | 5,388 | -35% | 1 | 1 | 0% | 1,492 | 1,719 | +15% | 0 | 0 | — |
case-22 | pass→pass | 12,061 | 9,929 | -18% | 1 | 1 | 0% | 1,846 | 2,560 | +39% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 19 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +50 percentage points is the difference between those two pass rates over the 19 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.