Connect instructions, a model, and tools into a local development loop you can inspect.
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.
The worked example is illustrative and has not been tested in a signed-in product account.
What it does and how it connects
Connect instructions, a model, and tools into a local development loop you can inspect.
Cursor’s Agent combines instructions, tools, and a selected model. It can search, edit, and run commands in an allowed development loop. Projects can coordinate broader work, while checkpoints capture agent file changes. A checkpoint is not the same as a repository commit.
Walkthrough
Use the product named in the access panel. The steps below combine documented controls with the editorial example.
- Open a sample repository and inspect existing guidance and working-tree changes.
- Ask Agent to explain a narrowly defined problem before editing.
- Review proposed commands and resulting file changes.
- Run the relevant checks and preserve a deliberate Git checkpoint before further work.
Worked example
Fix a sample mobile navigation overflow with a clear width and expected behavior. Ask the agent to inspect the layout and verify the reported viewport. Preserve unrelated styling so the review stays understandable.
In this sample repo, fix the navigation overflow at [viewport]. Inspect the layout first. Keep the visual system and ordinary document links. Verify narrow and wide layouts and report actual checks. Preserve unrelated changes; do not push or deploy.
Inspect and verify the result
Inspect the diff and test the viewport and keyboard path. Check tool output for actual verification. Confirm a change to an allowed config did not inadvertently broaden runtime authority.
Limits and troubleshooting
An agent needs the right context and a working environment to verify a fix. Checkpoints track agent changes rather than all manual work and conversation state. Use Git for deliberate version history.
Data, permissions, and actions
Agent context can include code, files, and tool results. Local edits can trigger live reload or execute changed code. Review commands and network access and keep production secrets outside a learning task.
Related reading
Rules and AGENTS.md
Scope reusable project guidance and inspect when it applies.
A local UI repair with deliberate recovery
Fix a sample UI problem while preserving unrelated work.
A cloud change with source and environment review
Run a scoped remote task and hand off tested branches or pull requests.
Evaluations: does the system do the job?
Create representative tests for quality, tool behavior, refusal, escalation, and the effects of a change.
Sources and verification
Documentation checked 2026-10-04. Examples are illustrative unless described otherwise.
Agent fundamentals and checkpoints · Cursor