---

name: whisper-lolo-roadmap description: Guide development of the whisper-lolo project based on specifications-projet.md. Use when planning or executing a sprint/PR, validating scope or constraints, or aligning architecture, statuses, and DoD for the Next.js + Vercel + Blob + Inngest + Whisper stack.

Whisper Lolo Roadmap

Overview

Apply the project roadmap and constraints to keep each PR aligned with the sprint goals, status model, and non-negotiable rules.

Workflow

1) Identify the sprint and scope

• Confirm which sprint (0..5) the task targets.
• Keep the PR limited to one sprint and its Definition of Done.
• When unsure, ask which sprint the change belongs to.

2) Re-assert hard constraints

• Never upload audio via a Next.js API route.
• Never wait for transcription inside an HTTP request.
• Always chunk, store, upload direct, then process async.
• Chunk before transcription.

3) Maintain data model and statuses

• Use recordings.status: draft | recording | uploaded | transcribing | done | error.
• Use transcriptions.status: pending | done | error.
• Update status transitions explicitly in code paths.

4) Apply sprint DoD checks

• Sprint 0: migrations applied, endpoint test ok, /record reachable.
• Sprint 1: 10 minutes stable, refresh restores chunks, blob playable.
• Sprint 2: long upload ok, Blob URL reachable, DB updated.
• Sprint 3: event received, job starts automatically.
• Sprint 4: transcription stored, errors handled, logs clear.
• Sprint 5: status shown, text rendered, copy/export works.

5) Reference project docs

• Read specifications-projet.md for the authoritative roadmap.
• Use documentation/ summaries to confirm API behaviors.

Key decisions to preserve

• Next.js App Router + TypeScript.
• Inngest for long-running jobs.
• Vercel Blob client uploads.
• Provider abstraction for Whisper now, Voxtral later.

References to consult

• specifications-projet.md
• documentation/vercel-blob-overview.md
• documentation/vercel-blob-sdk.md
• documentation/inngest-background-jobs.md
• documentation/openai-node-audio-example.md