openclaw - 💡(How to fix) Fix Assistant text output is auto-sent to channel even when tool calls follow [1 participants]

TL;DR

Setting agents.<id>.directives.autoReply=false may prevent automatic text-to-channel delivery and mitigate the 'narration leak' issue.

Guidance

• Verify the current behavior by checking if text portions of assistant responses are being sent as independent messages to the channel.
• Test the suggested fix by adding agents.<id>.directives.autoReply=false to the configuration and observing if the issue is resolved.
• Consider implementing a mechanism to send the final assistant text only after all tool calls are complete, as an alternative solution.
• Review the skill-level rule and AGENTS.md documentation to ensure consistency with the proposed fix.

Example

No code snippet is provided as the issue does not contain specific code references.

Notes

The effectiveness of the suggested fix may depend on the specific implementation and version of the system, which is not specified in the issue.

Recommendation

Apply workaround: Setting agents.<id>.directives.autoReply=false is a straightforward and targeted solution to prevent automatic text-to-channel delivery and address the 'narration leak' issue.