# S3 Archive Foundations Runbook

This runbook documents the currently implemented storage/archive foundations and the boundaries of current behavior.

## Implemented Now

- config modeling for:
  - `pipeline.storage.s3`
  - `pipeline.spool`
  - `pipeline.archive`
  - `session.inputs.audio_s3`
- promotion rule validation for safe relative paths
- run ID generation and path/key helper functions
- manifest run/path identity fields

## Not Implemented Yet

- real S3 backend integration
- AWS SDK wiring
- prepare-stage S3 object listing or download
- archive-stage S3 upload or promotion writes
- writing `current/manifest.json` or `current/run_id.txt` in S3

## Operational Notes

- local audio workflows remain the active development path (`audio_dir` or `audio_files`)
- `audio_s3` and local audio config are mutually exclusive
- do not place AWS credentials in Narratio config files

## Next Implementation Target

Build the storage backend layer that can list/download/upload S3 objects through a fake-tested interface, then wire prepare/archive stages to use that backend.