Background
I’ve observed that in a Copilot session, there is built-in context summarization behavior that helps manage long conversations and large context.
However, I did not observe similar behavior when using subagents or custom agents.
Question
Do subagents or custom agents have any context management capabilities, such as:
- Automatic context summarization
- Context truncation or compression
- Memory management across long-running tasks
Motivation
I’m evaluating whether it is suitable to assign long-running or large-context tasks to subagents/custom agents, and whether they can handle context growth reliably.
Additional Context
If this capability is not currently supported, are there any recommended patterns or best practices for handling large context in subagents?
Thanks!
Subagent docs: https://docs.github.com/en/copilot/concepts/agents/copilot-cli/comparing-cli-features#subagents
Background
I’ve observed that in a Copilot session, there is built-in context summarization behavior that helps manage long conversations and large context.
However, I did not observe similar behavior when using subagents or custom agents.
Question
Do subagents or custom agents have any context management capabilities, such as:
Motivation
I’m evaluating whether it is suitable to assign long-running or large-context tasks to subagents/custom agents, and whether they can handle context growth reliably.
Additional Context
If this capability is not currently supported, are there any recommended patterns or best practices for handling large context in subagents?
Thanks!
Subagent docs: https://docs.github.com/en/copilot/concepts/agents/copilot-cli/comparing-cli-features#subagents