Alchemy IDE is an experimental MIGI satellite project for authoring explainable, permission-aware workflows.
Its intended role is to help developers and creators design workflows that connect symbolic intent, typed schemas, policy checks, tools, models, receipts, and user interfaces.
This repository is a development workspace. It does not claim that every planned capability is implemented or production-ready.
MIGI Core defines the authority and provenance boundaries:
Original → Observation → Authority → Execution → Receipt
Alchemy IDE is intended to make those workflows visible and editable:
Symbolic intent
↓
Typed workflow graph
↓
Policy validation
↓
Tool / model / service execution
↓
MUEF event + MIGIReceipt
- visual workflow authoring;
- Emoji A++ or symbolic command parsing;
- typed event and receipt schemas;
- policy linting before execution;
- tool and model capability registry;
- replayable execution graphs;
- debugging views for authority and provenance;
- adapters that keep external services behind explicit permission boundaries.
- Do not represent simulations as verified observations.
- Keep policy evaluation separate from model inference.
- Require explicit authority for consequential actions.
- Preserve third-party licenses and attribution for dependencies or borrowed code.
- Link durable execution changes to MUEF events and MIGIReceipt records where applicable.
Use these labels in documentation and issues:
implemented
prototype
planned
research
upstream dependency
Architecture, shared trust primitives, and the MIRROR SEED MVP backlog live in QSTARMIGI/MIGI.