Ability to intercept or emulate agent messages in plugins

[yurihbm]

Question

I would like to know if there is a recommended way to:

• Inject Agent Responses: Send a message that is visually and functionally treated as an "Assistant/Agent" output (including markdown/syntax highlighting).
• Message Interception: A hook or middleware to intercept a message generated by the agent before it reaches the UI, allowing for real-time transformation.

Context

I am developing a plugin where I need to either intercept agent messages for transformation or send messages on behalf of the agent.

Current Behavior

Currently, using client.session.prompt with the noReply: true option allows sending a message to the session, but it has two main limitations for this use case:

• User Role: The message is sent as the user, not the agent.
• Formatting & Costs: These messages appear to consume input tokens and lack the syntax highlighting typically applied to agent outputs.

[github-actions]

Thanks for updating the issue! It now follows the contributing guidelines and the needs:compliance label has been removed.