Prompt Cards

ChatGPT drafts prompts and adversarial tests; Cursor integrates an eval harness; Antigravity schedules nightly evals and posts regressions with diffs. Include versioning and approval flow.

Pick a UI/server framework migration (e.g., Vue→React, Express→FastAPI). ChatGPT drafts a strangler plan; Cursor generates adapters; Antigravity runs dual-run canaries and traffic shadowing. Output risk/mitigation grid.

ChatGPT converts commits/PRs into user-facing notes, Cursor wires scripts into release pipelines, Antigravity tags versions and posts notes to docs/Slack. Include deprecation warnings template.

ChatGPT generates robust Playwright/Cypress specs with retry/wait heuristics; Cursor adds data-seeding helpers; Antigravity shards and retries, quarantines flaky tests, and files deflakes. Provide a flake budget.

ChatGPT codifies component usage rules, Cursor writes lint/codemods to enforce tokens/spacing, Antigravity gates CI on violations and auto-opens refactor PRs. Output exemptions policy.

Use Cursor’s visual editor to propose UI tweaks, ChatGPT generates design-token updates and a11y copy, Antigravity snapshots and runs visual diffs to ensure no unintended changes. Deliver a review pack.

For a data/ETL repo, ChatGPT drafts data quality tests (nulls, uniqueness, freshness), Cursor adds dbt/unit tests, and Antigravity schedules validations on PR. Include lineage notes and rollback triggers.

Define performance budgets and benchmarks; ChatGPT proposes profiling plan; Cursor instruments code (e.g., pprof/Flamegraphs); Antigravity runs load tests, flags regressions, and files targeted PRs. Output a perf dashboard spec.

Collect logs/incidents; ChatGPT drafts a blameless postmortem; Cursor queries log/trace snippets and links to code; Antigravity reconstructs a timeline and verifies action items land in code/config. Provide a 30-minute weekly review ritual.

Draft a PR reviewer flow: ChatGPT summarizes diffs and intent; Cursor leaves inline suggestions and auto-fix commits; Antigravity validates tests and risk scores before merge. Output a reviewer rubric and merge checklist.

Create a lightweight SDL: ChatGPT runs threat modeling (STRIDE) per feature, Cursor integrates SAST/secret scan hooks, Antigravity executes dependency/SBOM scans and opens remediation PRs. Provide severity ladder and SLAs.

ChatGPT authors OpenAPI/AsyncAPI contracts; Cursor code-gens server stubs/SDKs and adds contract tests; Antigravity spins up mocks and runs compatibility checks. Output versioning policy and deprecation calendar.