Issue with agents
I have multiple issues with the agents, making them unusable. I need you to guide me on fixing these issues because I'm struggling with OpenClaw and feel like my invested time is getting nowhere.
I've provided a list of skills for my agents, which were installed from the Claw Hub. I need you to consider these skills and evaluate if they're causing the problems. The main issues are:
* Agents don't remember their previous actions. For example, when I ask them to continue a task from the morning in the afternoon, they reply that they can't because they need an API to search, even though they're connected to the LLM and have access to their memory file.
* Both main and sub-agents have the same issue when asked to recall previous tasks. They claim to need an API to search their own memory, which doesn't make sense since they're connected to the LLM.
* Agents frequently stop working on tasks. For instance, I asked an agent to browse an online guide, parse information, and download it to an MD file. It would download a few guides, stop, and give an empty excuse when asked why. Restarting the task would result in it downloading a few more guides, but it wouldn't complete the task.
I need you to directly assist my agent. Please perform an audit and ask the agent to provide the findings under 2000 characters with sanitized tags for privacy. After receiving the agent's response, I'll copy and paste it back to you. You'll then use this information to help resolve the ongoing issues with the agent's functionality. I'll provide additional information about the installed skills in my next message. Remember that you're directly commanding my agent. Don't command me, just write a prompt directly to my agent that I can copy easily from a text box.
I've provided a list of skills for my agents, which were installed from the Claw Hub. I need you to consider these skills and evaluate if they're causing the problems. The main issues are:
* Agents don't remember their previous actions. For example, when I ask them to continue a task from the morning in the afternoon, they reply that they can't because they need an API to search, even though they're connected to the LLM and have access to their memory file.
* Both main and sub-agents have the same issue when asked to recall previous tasks. They claim to need an API to search their own memory, which doesn't make sense since they're connected to the LLM.
* Agents frequently stop working on tasks. For instance, I asked an agent to browse an online guide, parse information, and download it to an MD file. It would download a few guides, stop, and give an empty excuse when asked why. Restarting the task would result in it downloading a few more guides, but it wouldn't complete the task.
I need you to directly assist my agent. Please perform an audit and ask the agent to provide the findings under 2000 characters with sanitized tags for privacy. After receiving the agent's response, I'll copy and paste it back to you. You'll then use this information to help resolve the ongoing issues with the agent's functionality. I'll provide additional information about the installed skills in my next message. Remember that you're directly commanding my agent. Don't command me, just write a prompt directly to my agent that I can copy easily from a text box.
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
131,281Members
Resources
Similar Threads
Was this page helpful?
