Use an interactive agent for a bounded multi-file change and review each consequential command.
Coding & development · Documented · Documentation checked 2026-10-04
Editor assistance and agents, GitHub cloud agent, and code review have different execution contexts.
Plan, supported IDE, repository permission, AI credits, and organization policy determine access.
The worked example is illustrative and has not been tested in a signed-in product account.
What it does and how it connects
Use an interactive agent for a bounded multi-file change and review each consequential command.
IDE agent mode can select files, edit them, propose commands, and iterate. It remains interactive: you can redirect, review changes, and approve commands according to editor policy. This is different from GitHub’s cloud agent working on a repository task.
Walkthrough
Use the product named in the access panel. The steps below combine documented controls with the editorial example.
- Open Copilot Chat in the supported IDE and select available Agent mode.
- Give a reproducible issue and a scoped expected result.
- Review streamed edits and proposed commands before allowing them.
- Run appropriate checks and inspect the final diff.
Worked example
Fix a sample page whose search result count disagrees with the visible results. Supply the input and expected count. Ask the agent to inspect the data filter and render path together, then verify both normal and empty cases.
Fix this sample search-count mismatch: [reproduction]. Inspect the relevant filter and rendering code, preserve unrelated edits, and verify normal and empty-result cases. Report actual command outcomes. Do not push, merge, or change external services.
Inspect and verify the result
Compare result rows with the count in the rendered or tested behavior. Inspect changed files and command output. Check editor settings if commands ran automatically rather than assuming a universal approval default.
Limits and troubleshooting
Agent mode can be disabled by an administrator. Supported IDE steps differ. Missing dependencies can prevent checks; report that blocker. Each agentic request uses the applicable GitHub AI credit budget.
Data, permissions, and actions
Agent mode can change local files and execute commands. MCP may add external tools. Keep the repo and network scope narrow and review consequential operations beyond ordinary local edits.
Related reading
Instructions, custom agents, and MCP
Combine repository guidance with explicitly authorized tools, without treating instructions as security controls.
An editor fix with a checked handoff
Move from a checked suggestion to a scoped IDE agent change.
From a clear issue to a reviewed pull request
Create a cloud-agent change and a human-reviewed pull-request handoff.
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.
Use agent mode in your IDE · GitHub