Skip to content

Simplify SQLite config file name mangling for custom models #226

Description

@exedev-shelley

Author: philz42
Channel: #feedback
Link: https://discord.com/channels/1405685085923049482/1405686161791516873/1517200799455772883


Summary

The SQLite config file uses name mangling for custom models (e.g., custom-{mangle}), which makes it harder to reference these models in system prompts and configuration. Need to simplify this mangling.

Context

From discussion about idempotent model configuration:

warpfork: It would be nice to have an idempotent API for setting available models. Or a config file.

I also have a bunch of custom agents and am a little frustrated at how much friction it is to both load those up on new VMs, and to make sure the default ones aren't selected. (I do have guidance in my system prompt about the choice, but it only helps for the latter part, and somewhat probabilistically.)

The current ui that results in the API name being "custom-{mangle}" also makes it a tad harder to talk about in my system prompt.

Notes

  • The SQLite file serves as the config file
  • This affects the ability to reliably reference custom models in system prompts and across VM deployments
  • Related to broader issue of having an idempotent config mechanism for models and custom agents

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions