run_in_terminal: detect shell continuation prompts and abort#316271
Closed
meganrogge wants to merge 2 commits into
Closed
run_in_terminal: detect shell continuation prompts and abort#316271meganrogge wants to merge 2 commits into
meganrogge wants to merge 2 commits into
Conversation
When a command with unmatched quotes is sent to the terminal, the shell enters a continuation prompt (e.g. dquote>, quote>) and waits for input. Previously this caused the terminal tool to hang indefinitely because no command-finished event ever fires. This change: - Adds isContinuationPrompt() to detect known zsh continuation prompts - Adds waitForContinuationPrompt() that monitors terminal idle state for continuation prompt patterns - Integrates detection as a race candidate in all three execute strategies (rich, basic, none) - When detected, sends Ctrl+C to abort and returns an error message explaining the quoting issue - Fixes detectsCommonPromptPattern() to exclude continuation prompts so they are not falsely detected as ready prompts Fixes #315694 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Contributor
There was a problem hiding this comment.
Pull request overview
Fixes a hang in the run_in_terminal tool when commands contain unmatched quotes/brackets and the shell drops into a continuation prompt (e.g. dquote>). All three execute strategies now race a new continuation-prompt detector alongside their existing completion signals, send Ctrl+C when one is seen, and report a syntax-error message back to the agent. detectsCommonPromptPattern is also fixed so it no longer mistakes continuation prompts for ready prompts.
Changes:
- Add
isContinuationPromptandwaitForContinuationPrompthelpers and exclude continuation prompts fromdetectsCommonPromptPattern. - Race continuation-prompt detection in
RichExecuteStrategy,BasicExecuteStrategy, andNoneExecuteStrategy; on detection send Ctrl+C and return an explanatory error. - Add unit tests for
isContinuationPromptand for the newdetectsCommonPromptPatternrejection cases.
Show a summary per file
| File | Description |
|---|---|
| src/vs/workbench/contrib/terminalContrib/chatAgentTools/browser/executeStrategy/executeStrategy.ts | Adds continuation-prompt patterns/helpers and excludes them from common prompt detection. |
| src/vs/workbench/contrib/terminalContrib/chatAgentTools/browser/executeStrategy/richExecuteStrategy.ts | Races continuation-prompt detection and aborts with Ctrl+C on detection. |
| src/vs/workbench/contrib/terminalContrib/chatAgentTools/browser/executeStrategy/basicExecuteStrategy.ts | Same continuation-prompt race + abort handling for the basic strategy. |
| src/vs/workbench/contrib/terminalContrib/chatAgentTools/browser/executeStrategy/noneExecuteStrategy.ts | Adds continuation-prompt branch to the prompt/alt-buffer race; Ctrl+C call is not awaited unlike sibling strategies. |
| src/vs/workbench/contrib/terminalContrib/chatAgentTools/test/browser/executeStrategy.test.ts | Tests for isContinuationPrompt and the new rejection cases in detectsCommonPromptPattern. |
Copilot's findings
- Files reviewed: 5/5 changed files
- Comments generated: 2
…prompt detection
- Await sendText('\x03') in noneExecuteStrategy for consistency with
rich and basic strategies
- Use 2x idle interval for continuation prompt detection to reduce
false positives from commands that briefly print prompt-like text
- Add cursor column check to verify the shell is actually awaiting
input at the prompt, not mid-line in command output
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
dmitrivMS
approved these changes
May 13, 2026
Collaborator
Author
|
/requires-eval-assessment terminalbench2 gpt-5.4,claude-opus-4.6,claude-opus-4.7 |
meganrogge
added
the
~requires-eval-assessment
Contributor
|
⏳ Queued vscode build for
|
Collaborator
Author
|
I actually think this is overkill / too narrowly scoped and will be fixed by #315885 |
Contributor
|
🚀 Queued eval-assessment publish build for
|
Contributor
|
❌ Eval-assessment build did not succeed (result: |
vs-code-engineering
Bot
removed
the
~requires-eval-assessment
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #315694
When a command with unmatched quotes is sent to the terminal, the shell enters a continuation prompt (e.g.
dquote>,quote>) and waits for more input. Previously this caused the terminal tool to hang indefinitely because no command-finished event ever fires.Changes
executeStrategy.ts:isContinuationPrompt()to detect known shell continuation prompts (dquote>,quote>,bquote>,pipe>,cmdsubst>,heredoc>)waitForContinuationPrompt()that monitors terminal idle state and checks for continuation prompt patterns on the cursor linedetectsCommonPromptPattern()to exclude continuation prompts so they are not falsely detected as ready prompts (previouslydquote>matched the generic[>%]pattern)All three execute strategies (rich, basic, none):
waitForContinuationPrompt()as a race candidate alongside existing completion signalsHow it works
The fix monitors the terminal data stream for idle periods. When the terminal goes idle, it checks if the cursor line matches a known continuation prompt pattern. If so, it:
This prevents the hang and gives the agent actionable feedback to fix the command.