SemantiFX

Text prompt in, editable effect-chain configuration out. A JUCE plugin (plugin/, built and iterated on macOS) talks to a Python inference worker (inference/, developed on Linux with CUDA) over a JSONL protocol. Chain optimization tunes a fixed chain's parameters with a CLAP loss. Chain assembly picks and orders the effects with TypeSafe Jev.

Read CONTEXT.md for the vocabulary before naming anything.

Working here

• Solo repo. Work directly on main; no pull requests.
• Python: cd inference && .venv/bin/python -m unittest discover -s tests. Set SEMANTIFX_RUN_MODEL_TESTS=1 to include the golden CLAP test.
• API keys live in .env at the repo root (gitignored). Never print them.
• Experiment write-ups go in docs/experiments/.

Agent skills

Issue tracker

GitHub Issues on flamboh/semantifx, driven with gh. See docs/agents/issue-tracker.md.

Triage labels

Default vocabulary (needs-triage, needs-info, ready-for-agent, ready-for-human, wontfix). See docs/agents/triage-labels.md.

Domain docs

Single-context: CONTEXT.md at the root, ADRs under docs/adr/. See docs/agents/domain.md.