Copilot Agent waits after a terminal command? Verify its exit result.

If the terminal displays a completed command while Copilot Agent keeps waiting, inspect the actual process and terminal result before sending another instruction.

Do this first: Confirm the command has an exit result and is no longer running. Preserve terminal output and the diff, then avoid pressing Enter or launching another task until you know what state the agent received.

What the failure looks like.

The terminal appears to finish a command, but the Agent run remains waiting or spinning. Issue #13592 is one historical report of this terminal/agent state mismatch.

Why it happens.

A displayed line of output is not always a process exit signal. The issue records one environment-specific report; it does not establish that PowerShell, line endings, or Copilot generally cause this symptom.

If Copilot is waiting for approval.

A permission or approval deadlock is different from a finished terminal command. Preserve the diff, identify the exact approval prompt, then restart with a smaller command or explicit permission boundary. Related failures: Copilot infinite retry loop and Copilot send button disabled.

Fix it without losing the task.

Check the process list or terminal prompt for a real command completion. Capture the exit code and full output.

Inspect Copilot Chat and terminal separately. Record whether the Agent received the output or remains on the same step.

Save the working tree diff before stopping or restarting the Agent. Avoid sending a second command until the first process is accounted for.

Resume with the terminal result and one explicit verification step. If the files fail that check, switch to repository-based validation rather than repeating the command.

What the next AI should receive.

VS Code and Copilot versions, terminal profile, exact command and exit status, whether a process remained active, output visible to the Agent, current diff, and next verification.

First safe action

Check for a real exit code or a still-running child process. Save the exact command and terminal surface before stopping or restarting the Agent.

Evidence boundary

Issue #13592 concerns Copilot Chat 0.28.2 and VS Code 1.101.2 and is closed as a duplicate. It documents a reported wait after terminal output, not a current general defect.

Diagnostic checkpoint

This page is for an Agent run still waiting after a terminal command. If the Agent says it is complete but the files or checks are wrong, use the separate repository-verification guide.

How ShardStitch helps.

ShardStitch can preserve the task goal, changed files, terminal output, and last verification for a fresh session. It cannot relay a missed terminal exit event or determine whether a command is still active.

Source trail.

Copilot issue #13592 and the Copilot issue tracker.

FAQ.

Does visible terminal output prove Copilot received the command result?

No. Confirm the process exit and inspect the Agent state separately. The user-visible terminal and agent tool state can diverge.

Should I press Enter in the terminal to wake Copilot?

That was a reporter-observed action in one historical issue, not a verified general workaround. Preserve the current state and first confirm whether the command is still active.

Scope note: This is a terminal-state diagnostic anchored to one historical Copilot report. It is not evidence that completion indicators or all terminal integrations are unreliable.