Skip to content

Certify personal capture audio pipeline - #1

Draft
Pepstee wants to merge 17 commits into
mainfrom
codex/personal-capture-v0
Draft

Certify personal capture audio pipeline#1
Pepstee wants to merge 17 commits into
mainfrom
codex/personal-capture-v0

Conversation

@Pepstee

@Pepstee Pepstee commented Jul 30, 2026

Copy link
Copy Markdown
Owner

What changed

  • reorganizes the repository around explicit CLI, reusable ML engines, web, and Personal Capture application boundaries
  • hardens long-audio ingestion with resumable transcription, diarization and translation stages
  • adds offline model pinning and whole-file hash manifests, host preflight, private review flow, and transactional GIGA outbox handling
  • aligns deployment assets, documentation, schemas, and tests with the production folder layout
  • includes the forensic archaeology reports and remediation evidence while excluding generated audit scratch environments

Why

The first real full-day recording must enter a clean, testable, fail-closed environment rather than exposing unresolved architecture, path, provenance, security, or recovery decisions during production use.

Validation

  • 802 tests passed, 2 skipped
  • Python compileall passed
  • Ruff passed
  • wheel and source distribution builds passed
  • clean-wheel console entrypoint checks passed
  • staged diff whitespace check passed

Host certification remains a separate deployment gate: the exact commit will be fetched on the Gigabyte, the pinned local models will be manifested and loaded offline, and dubbing-capture-preflight --prepare --load-models must report ready before real audio is admitted.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant