Current plugin setup makes development/debugging very hard

[samlaf]

The main use case for plugins which we can see (and that we currently need!) is to make a cli for interacting with the avs contracts.

Our current setup which we are trying to migrate to using this cli is to start a local anvil chain that is forking goerli, and then use the cli to:

• deploy avs contracts
• register with those contracts
• make other useful calls (stake updates, force undelegate staker, deregister, etc.)

Here's our example repo which we are currently working off of.

Currently it seems like plugins are just run in a docker container on the same network as the profile. Few things which would be nice to have:

• be able to run plugin container on host network (so that it can point to the anvil chain at localhost:8545)
• be able to mount a local directory so that cli outputs can persist state to future cli calls (this might even be necessary in prod!)
• be able to run plugin calls without needing to have the monitoring stack + avs pkg code running (might be useful to make certain read calls to the contracts before installing/running the offchain code for eg)

[AntiD2ta]

Thanks for the detailed feedback, those points make sense.

[AntiD2ta]

be able to run plugin calls without needing to have the monitoring stack + avs pkg code running (might be useful to make certain read calls to the contracts before installing/running the offchain code for eg)

Regarding this, it will require some re-design of the plugin system to be able to run it without the avs pkg code running but as long as the avs pkg is installed then should be doable. By default the avs is executed after being installed but we can make it configurable for an install only.

Not running the monitoring stack is an issue for the system and future plans and I would love to discuss this better.