Fix a sample UI problem while preserving unrelated work.
Coding & development · Documented · Documentation checked 2026-10-04
Cursor editor agents and Cloud Agents; Grok Bot has a separate shared-computer boundary.
Installed client, subscription, organization settings, repository access, and cloud environment determine capabilities.
An illustrative recipe. This website does not run the automation.
The goal and the right kind of system
Fix a sample UI problem while preserving unrelated work.
An interactive Agent can inspect and iterate. A person checks the visual / behavioral result and keeps Git recovery separate from agent checkpoints.
Inputs and connected services
Inputs: sample repo, reported viewport / state, existing rules, and working-tree checkpoint.
Services: local Cursor Agent and available project checks.
The steps
An illustrative sequence to adapt and test before using real data or enabling ongoing work.
- Inspect rules and the relevant rendering path.
- Make one scoped correction with reviewed commands.
- Check the specified UI state and wide / narrow behavior.
- Read the diff and preserve a deliberate Git checkpoint.
Outputs and human checkpoints
Outputs: scoped code change and verified / unverified behavior.
Human checkpoints: review changes and any proposed restore before applying it.
Exceptions, permissions, and stopping conditions
Permission boundary: no external writes. Live reload may execute edits, so use a sample environment.
Stop on unrelated modifications, unverified restoration scope, failing checks, or unclear context.
Adaptable prompt
Fix [sample UI state] while preserving the design and unrelated edits. Verify the specified viewports, report actual checks, and show the diff. No push or deployment; do not restore over human changes without review.
Related reading
Agent, codebase context, and checkpoints
Connect instructions, a model, and tools into a local development loop you can inspect.
Rules and AGENTS.md
Scope reusable project guidance and inspect when it applies.
Review, recovery, and operational limits
Use diffs, checks, and checkpoints as complementary evidence rather than automatic acceptance.
Designing an AI workflow
Make a useful task repeatable with clear inputs, steps, outputs, exceptions, and human checkpoints.
Permissions and meaningful approvals
Decide what a system may read, what it may change, and when someone needs to authorize a concrete action.
Sources and verification
Documentation checked 2026-10-04. Examples are illustrative unless described otherwise.
Agent fundamentals and checkpoints · Cursor
Rules and AGENTS.md · Cursor
Agent security and approvals · Cursor
Cloud Agents and environments · Cursor