[GH-ISSUE #126] does this work with LM Studio Bionic? #28

New issue

Closed

opened 2026-10-15 20:52:37 +00:00 by bsakaguchi · 2 comments

bsakaguchi commented

2026-10-15 20:52:37 +00:00

Owner

Originally created by @Raskudzas on GitHub (Oct 6, 2026).
Original GitHub issue: https://github.com/rehan-remade/universal-modder/issues/126

or should we use something else for local AI agents?

Originally created by @Raskudzas on GitHub (Oct 6, 2026). Original GitHub issue: https://github.com/rehan-remade/universal-modder/issues/126 or should we use something else for local AI agents?

bsakaguchi closed this issue

2026-10-15 20:52:38 +00:00

bsakaguchi commented

2026-10-15 20:52:38 +00:00

Author

Owner

@rehan-remade commented on GitHub (Oct 7, 2026):

It should mostly work, though we haven't tested Bionic ourselves. Bionic supports the Agent Skills format (SKILL.md), which is what this repo's skills use. You can install a skill from a GitHub URL, or add a SKILL.md under Settings → Skills (docs). Start with skills/mod-any-game and skills/game-recon.

Two things we haven't checked: whether Bionic reads AGENTS.md, and whether it can connect to the fal MCP server. If it can't, the um CLI (um fal ...) does the same asset generation from a shell.

If you'd rather use a setup we have tested: OpenCode can use LM Studio's local server as its model provider, and inside a clone it loads this repo's skills and the fal MCP server through opencode.json.

Either way, the model matters more than the app. Modding sessions are long and make a lot of tool calls, so use the strongest open model you can run, or Bionic's cloud models for the hard parts. Small local models tend to lose track.

@rehan-remade commented on GitHub (Oct 7, 2026): It should mostly work, though we haven't tested Bionic ourselves. Bionic supports the Agent Skills format (`SKILL.md`), which is what this repo's skills use. You can install a skill from a GitHub URL, or add a `SKILL.md` under **Settings → Skills** ([docs](https://lmstudio.ai/docs/bionic/agent/skills)). Start with `skills/mod-any-game` and `skills/game-recon`. Two things we haven't checked: whether Bionic reads `AGENTS.md`, and whether it can connect to the fal MCP server. If it can't, the `um` CLI (`um fal ...`) does the same asset generation from a shell. If you'd rather use a setup we have tested: OpenCode can use LM Studio's local server as its model provider, and inside a clone it loads this repo's skills and the fal MCP server through `opencode.json`. Either way, the model matters more than the app. Modding sessions are long and make a lot of tool calls, so use the strongest open model you can run, or Bionic's cloud models for the hard parts. Small local models tend to lose track.

bsakaguchi commented

2026-10-15 20:52:38 +00:00

Author

Owner

@rehan-remade commented on GitHub (Oct 10, 2026):

Closing, since this has been answered above. If you get stuck trying it, open a new issue with the game, what you tried and the last error.

@rehan-remade commented on GitHub (Oct 10, 2026): Closing, since this has been answered above. If you get stuck trying it, open a new issue with the game, what you tried and the last error.

No labels

pull-request

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

rehan-remade/universal-modder#28

No description provided.

Rows
Columns