core.Agent.from_workspace()

[ph-ausseil]

Duplicates

• [ ] I have searched the existing issues

Summary 💡

Hi,

As of now core.Agent is loaded using method from_workspace() which is very restrictive. First of all because there is hope to support various BDD (Redis, ...) & because workspace is semantically meant to store Agents artefacts & productions.

An agent should be loaded from a database would it be json_file or not ; an agent should not be loaded from a workspace.

I'm reporting this as an issue for due tracking and quote my Discord report over the issue :

What was to be read between the line is that may be memory-related classes should be passed additional elements that are not passed today ton the only implementation we have. These element may include

• a table name/ filename ,
• a key,
• a json/object
• or a kwarg if location/port/credentials are needed.

Memory adapters should push informations to the relevant location & table .

[...]

As for json_file location, I suggest ~/autogpt/data/ would be much better and then we would have 3 files message_history, agents & users. State of the art would be to make the location fully configurable so if needed a user could host AutoGPT in the cloud but his data on premise

Examples 🌈

No response

Motivation 🔦

• The loading of agent in the rearch is buggy and it would be great to get it actually good
• The curent situation is limiting capacity to have 2 agents
• Would enable some work on API

[github-actions[bot]]

This issue has automatically been marked as stale because it has not had any activity in the last 50 days. You can unstale it by commenting or removing the label. Otherwise, this issue will be closed in 10 days.

[github-actions[bot]]

This issue was closed automatically because it has been stale for 10 days with no activity.