tone-agent-skill

This repo contains a public tone-refinement skill for bass and guitar work with LLMs.

This repo is vibe-coded, but tested in real life.

The material here was shaped through actual interaction with GitHub Copilot in VS Code and GPT-5.x chat-agent style workflows, then checked against real tone work instead of staying purely theoretical.

It is meant for practical tone engineering, not vague gear talk. The focus is:

• real routing constraints
• repeatable A/B workflow
• sample capture and measurement
• split-path design when useful
• IR selection and balancing
• validation in actual mix or live context

Main Files

SKILL.md

This is the actual skill file.

It contains the core reusable workflow for refining tones in:

• DAWs
• plugin chains
• digital modelers
• real rigs

prompts/copy-paste-tone-refinement.txt

This is a compact ready-to-use prompt for people who do not want to clone a repo or install anything.

The intended use is simple:

1. open ChatGPT, Gemini, or another LLM
2. paste the file
3. start working from there

References

The references/ directory contains supporting material used by the skill.

Current references include:

• sample capture and evaluation guidance
• IR selection workflow
• IR note templates
• bad-result diagnosis patterns
• platform tradeoffs
• a worked real-world bass example

These are not separate products. They support the main skill and give it deeper reusable context.

In the worked example, the resulting EB + DG paths were rendered and handed off to mixing for the to-be-released Enslave Them Stars song Your Future Is..., so the example is tied to a real release workflow and not only sandbox experimentation.

Repo Status

This is a public repo hosted on a private Forgejo instance.

Federation is not enabled yet.

Contributions

To submit patches, contact:

Tomasz Sieprawski tomasz@sieprawski.pl

Contributions are accepted only under the MIT license.