Monitor Baseline Before Big Refactors (/usage)
Run /usage or /quota inside the CLI/panel before launching multi-file autonomous tasks so you never unknowingly cross the weekly threshold.
Solve the notorious "5-hour vs 5-day trap": learn how Antigravity pairs short rolling sprint refills with an underlying weekly hard cap, and stop autonomous agent loops from burning your week.
Paid tiers (such as Google AI Pro) promote fast 5-hour rolling refreshes inside Antigravity. However, Google enforces an underlying 7-day weekly baseline cap beneath those sprint windows. If an intensive refactor, full-repo scan, or stuck retry loop burns your remaining weekly baseline inside one afternoon, the UI stops offering 5-hour resets and jumps straight to a 4–7 day lockout until your weekly cycle refreshes.
Run /usage or /quota inside the CLI/panel before launching multi-file autonomous tasks so you never unknowingly cross the weekly threshold.
Prevent context indexing from repeatedly ingesting node_modules, dist, .git, lockfiles, and binary assets on every agent step.
Reserve heavyweight Pro/Ultra reasoning models for architecture and debugging; keep CSS tweaks, boilerplate, and small fixes on Fast Mode.
If an agent repeatedly fails the same test or build error 2–3 times in a row, cancel the run manually and inspect the error yourself — infinite retry loops are the #1 cause of weekly burnouts.
# Dependencies & build artifacts
node_modules/
dist/
build/
.next/
.astro/
coverage/
# Version control & caches
.git/
*.log
*.lock
package-lock.json
pnpm-lock.yaml
# Large media & binary dumps
*.png
*.jpg
*.mp4
*.zip
*.sqliteEstimate when your 5-hour Gemini App / Antigravity window or 7-day weekly cap unlocks, or convert a CLI/UI remaining countdown into your exact local timestamp.