Fix compaction loop for small models or large input

[katzdave]

#5255

Track compaction attempts and give up if we get stuck in a doom loop (2 tries)

[Copilot]

Pull Request Overview

This PR fixes an infinite compaction loop that can occur when the context limit is exceeded for small models or with large inputs. The fix introduces a counter to track consecutive compaction attempts and stops after one failed retry with a helpful error message.

• Adds consecutive_compactions counter to prevent infinite loops when compaction doesn't resolve context limit errors
• Improves error message clarity in the compaction module
• Updates error handling to break instead of continuing when compaction fails

Reviewed Changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated 1 comment.

File	Description
ui/desktop/package-lock.json	Auto-generated npm dependency resolution changes (incidental, not related to fix)
crates/goose/src/context_mgmt/mod.rs	Improves error message when compaction fails after removing all tool responses
crates/goose/src/agents/agent.rs	Implements consecutive compaction counter and doom loop prevention logic

Files not reviewed (1)

• ui/desktop/package-lock.json: Language not supported

[Copilot]

The continue here continues the inner while let Some(next) = stream.next().await loop, but the stream was created before compaction (line 955). After successful compaction, you need to break from the inner loop so the outer loop (line 912) creates a new stream with the compacted conversation. Using continue here will just fetch the next item from the old stream, which won't use the compacted conversation.

Suggested change

	continue;
	break;

[katzdave]

fixed

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

[Copilot]

When compaction fails, the user no longer receives any feedback message. The old code yielded an AgentEvent::Message explaining the error and suggesting retry. Consider adding back a user-facing error message to inform them why the conversation stopped.

Suggested change

	error!("Compaction failed: {}", e);
	error!("Compaction failed: {}", e);
	yield AgentEvent::Message(
	Message::assistant().with_system_notification(
	SystemNotificationType::InlineMessage,
	"Unable to continue: Compaction failed due to an internal error. Try using a shorter message, a model with a larger context window, or start a new session."
	)
	);

[alexhancock]

Seems like a great thing to catch and handle!

[alexhancock]

I wonder if we could just check if the message length is a certain value? Conversation only contains a single user message and compaction failed?

This would be durable to other failure modes on the compaction request not working (like the provider connection is bad, or something else goes wrong)

What do you think?

[katzdave]

Most common situation this occurs is when pasting something gigantic, so I don't think the single message case is all that general.

That triggers out of context error -> triggers compaction -> loop.

Agree on the sentiment of seeing if the message length is a certain size, but a bit messy in practice and would need to both estimate token counts + have a good understanding of context window size and system prompt size.

I think will go with as is for now, but might be something there using the data from canonical models.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

[Copilot]

When compaction fails, no user-facing message is shown. The old code yielded an AgentEvent::Message explaining the error. Now users will see the conversation end without explanation, which is confusing. Consider adding back a user notification before the break.

Suggested change

	error!("Compaction failed: {}", e);
	error!("Compaction failed: {}", e);
	yield AgentEvent::Message(
	Message::assistant().with_system_notification(
	SystemNotificationType::InlineMessage,
	"Unable to continue: Compaction failed due to context limit. Try using a shorter message, a model with a larger context window, or start a new session."
	)
	);