Prompt the agent — watch it plan, run & self-correct
Goal. Write tasks the way agents work best — name the outcome and the check — so the agent can run and verify itself without you in the loop.
Refactor that UniProt script into a small module with type hints, then add a second function that fetches a sequence by accession. Add tests for both and run them — keep fixing until everything passes.
Send this as a follow-up in the same agent so it builds on the work already done instead of starting over. Use the mic in the prompt bar if you'd rather dictate.
- 1Give the agent a goal plus a way to check it — 'add tests and run them until they pass'. Because the agent can run code, a verifiable instruction lets it self-correct instead of guessing.
- 2Name what changes and where ('a second function that fetches a sequence by accession'). Specific, located asks beat 'make it better' — the same rule that helps any AI helps here.
- 3Watch the agent's plan and its run/fix loop in the Editor View — a full AI IDE with tab-completion and inline command requests right beside the running agent.
You'll see. The agent extending the earlier work in place — planning the change, writing code and tests, running them, and iterating on failures until the whole task is green.
Cost. Still on the free For Individuals plan; usage draws down against basic weekly rate limits, so batch related changes into one task rather than many tiny ones.
Takeaway. The best Antigravity prompts state an outcome and a check — give the agent something it can run and verify, and it closes the loop itself.