Moises

Build me a focused music services and creator audio tools workflow for the personal core of Moises. Requirements:

- Use Python 3.12, FastAPI, SQLite, ffmpeg, and a local web interface. Model owned tracks, file hashes, metadata, playlists or sessions, playback positions, and exports.
- Paid product context: Run a local stem-separation model and build practice controls for user-owned songs. Build only this DIY scope: Run a local stem-separation model on legally owned or user-created tracks, then provide private practice controls and playback for each song.
- For music services and creator audio tools, probe legally owned source tracks and retain their original hashes and rights metadata.
- Run the named local analysis or transformation as a job with settings and failure logs.
- Preview processed tracks and export new files plus a manifest of source, settings, and rights notes.
- Use a local web page with input, progress, review, and export views. Required input or access: legally owned local audio files; ffmpeg.
- Recovery: A failed render cannot replace the source; exported sample rate, channel count, and duration must match the manifest.
- Acceptance: with one labelled sample, show the input, saved intermediate state, and exported result; verify this invariant: A failed render cannot replace the source; exported sample rate, channel count, and duration must match the manifest.
- Out of scope: proprietary separation models, mobile polish, chord detection, cloud compute, and speed; licensed music catalog. Keep this a personal, inspectable workflow.
- Include a README with setup, a sample input, required keys or permissions, data location, and the supported scope.

# AGENTS.md — Build guide for Moises

## Project scope
Run a local stem-separation model on legally owned or user-created tracks, then provide private practice controls and playback for each song.

Catalogue verdict: no. A consolation build is possible, but the paid product's decisive value sits outside a solo rebuild. For Moises, run a local stem-separation model and build practice controls for user-owned songs. The hard boundary is proprietary separation models, mobile polish, chord detection, cloud compute, and speed, plus licensed content, rights, and distribution network.
Use the implementation prompt below to define the deliverable. Complete each phase's acceptance checks before extending the scope.

## Working agreement
- Inspect the repository and its existing instructions before choosing paths, dependencies or commands. Keep one coherent stack and explain changes to the proposed architecture.
- Plan a vertical slice that accepts a real input and produces the useful output described below. Persist only the state the prompt calls for; respect memory-only and upstream-managed workflows. Use fixtures only when they are clearly labelled.
- After scaffolding, document the actual install, development, check and build commands in README and keep them synchronized with the package or project manifest. Do not report commands as successful unless they ran.
- Work in small steps. At handoff, list implemented flows, checks actually performed, remaining blockers, and any credentials or provider setup the owner must supply.
- Do not publish, spend money, contact customers, delete source data or run irreversible migrations without the project owner's authorization.

## Prerequisites
- legally owned local audio files
- ffmpeg
- local storage
- optional audio-analysis models

## Stack and architecture
- Use Python 3.12, FastAPI, SQLite, ffmpeg, and a local web interface.
- Model owned tracks, file hashes, metadata, playlists or sessions, playback positions, and exports.
- Interface for this music services and creator audio tools workflow: a local web page with input, progress, review, and export views.

## Security and data integrity
- A failed render cannot replace the source; exported sample rate, channel count, and duration must match the manifest.
- Validate imported records and file paths; keep user data within the storage location named in the prompt.
- State the practical limit: proprietary separation models, mobile polish, chord detection, cloud compute, and speed.

## Agent implementation rules
- Project rule, data: Model owned tracks, file hashes, metadata, playlists or sessions, playback positions, and exports.
- Project rule, behavior: For music services and creator audio tools, probe legally owned source tracks and retain their original hashes and rights metadata.
- Project rule, recovery: A failed render cannot replace the source; exported sample rate, channel count, and duration must match the manifest.

## Optional agent skills and references
No installable skill is listed for this guide; use the project rules and the stack documentation.

Read the linked SKILL.md and its dependencies before adding a skill. Select only the skills matching this project's runtime and task; their documentation does not supply API access, credentials or approval to perform external actions. Pin the reviewed revision where the tool supports it. Follow the chosen agent's documented project-level installation mechanism.

## Distribution ideas
These are optional planning notes. Obtain the owner's approval before publishing or contacting anyone.
- Demonstrate this music services and creator audio tools operation with sample or user-owned data: For music services and creator audio tools, probe legally owned source tracks and retain their original hashes and rights metadata.
- Show a concrete result or export: Preview processed tracks and export new files plus a manifest of source, settings, and rights notes.
- State the practical limit: proprietary separation models, mobile polish, chord detection, cloud compute, and speed.

## Engineering roadmap
1. Phase 1, architecture and data: Use Python 3.12, FastAPI, SQLite, ffmpeg, and a local web interface. Model owned tracks, file hashes, metadata, playlists or sessions, playback positions, and exports.
2. Phase 2, implement: For music services and creator audio tools, probe legally owned source tracks and retain their original hashes and rights metadata.
3. Phase 3, implement: Run the named local analysis or transformation as a job with settings and failure logs.
4. Phase 4, review and output: Preview processed tracks and export new files plus a manifest of source, settings, and rights notes.
5. Phase 5, recovery and acceptance: Verify this invariant with a saved fixture: A failed render cannot replace the source; exported sample rate, channel count, and duration must match the manifest. State the practical limit: proprietary separation models, mobile polish, chord detection, cloud compute, and speed.

## Paid-product capabilities outside this build
- proprietary separation models, mobile polish, chord detection, cloud compute, and speed
- licensed music catalog
- royalty and distribution relationships
- global streaming infrastructure
- artist network and recommendations

## Implementation prompt
Build me a focused music services and creator audio tools workflow for the personal core of Moises. Requirements:

- Use Python 3.12, FastAPI, SQLite, ffmpeg, and a local web interface. Model owned tracks, file hashes, metadata, playlists or sessions, playback positions, and exports.
- Paid product context: Run a local stem-separation model and build practice controls for user-owned songs. Build only this DIY scope: Run a local stem-separation model on legally owned or user-created tracks, then provide private practice controls and playback for each song.
- For music services and creator audio tools, probe legally owned source tracks and retain their original hashes and rights metadata.
- Run the named local analysis or transformation as a job with settings and failure logs.
- Preview processed tracks and export new files plus a manifest of source, settings, and rights notes.
- Use a local web page with input, progress, review, and export views. Required input or access: legally owned local audio files; ffmpeg.
- Recovery: A failed render cannot replace the source; exported sample rate, channel count, and duration must match the manifest.
- Acceptance: with one labelled sample, show the input, saved intermediate state, and exported result; verify this invariant: A failed render cannot replace the source; exported sample rate, channel count, and duration must match the manifest.
- Out of scope: proprietary separation models, mobile polish, chord detection, cloud compute, and speed; licensed music catalog. Keep this a personal, inspectable workflow.
- Include a README with setup, a sample input, required keys or permissions, data location, and the supported scope.

## Completion evidence
Demonstrate the prompt's acceptance scenarios against the scoped workflow. Include setup from a clean checkout and failure recovery. Check persistence across restart and export/restore only for the state the prompt says to store; for memory-only tools, confirm that temporary content is discarded as specified. Record actual results and remaining limitations. A detailed plan alone does not establish a working replacement.