EchoPhrase Beta Review Plan - Dylagent & Elsagent Collaboration

echophrase-beta-review-plan.md

EchoPhrase Beta Review Plan

Goal: Comprehensive codebase review before beta launch
Participants: Dylagent (Lead Backend), Elsagent (Technical Director)

---

Scope Summary

Module	Tech Stack	LOC (approx)	Complexity
Client	Tauri + Rust + Svelte	~14k Rust, ~3k Svelte	High
Server	.NET 8	~30k (incl migrations)	Medium-High
Admin	React + TypeScript	~2k	Medium
WWW	SvelteKit	~1.5k	Low

---

Division of Labor

Dylagent (Lead Backend)

Focus: Client ↔ Server ↔ Admin integration flows

1. Client (Rust/Tauri)

   • Review large files for decomposition opportunities:
     • commands/transcription.rs (1717 lines) - candidate for split?
     • transcription/onnx_backend.rs (1087 lines)
     • transcription/candle_whisper.rs (1220 lines)
   • Feature gate implementation consistency
   • Auth flow (deep links, token storage, refresh)
   • Usage tracking / freemium enforcement
   • Entry point readability (lib.rs, command handlers)

2. Server (.NET)

   • Program.cs (840 lines) - break into extension methods?
   • Endpoint organization and auth middleware
   • Webhook handling (LemonSqueezy)
   • Entitlements service consistency

3. Client ↔ Server Integration

   • API contract alignment
   • Error handling consistency
   • Auth token flow end-to-end

4. Admin Dashboard

   • API client alignment with server
   • Feature gates editor sync with client/server
   • Constants editor validation

Elsagent (Technical Director)

Focus: WWW, cross-cutting concerns, architecture review

1. WWW (Marketing Site)

   • Pricing page accuracy vs actual tiers
   • Download page platform detection
   • i18n consistency
   • SEO metadata
   • Contact form integration

2. Cross-Module Feature Gates

   • Audit feature gate definitions across all modules
   • Ensure client UI matches server enforcement
   • Admin editor reflects all gates

3. Architecture Review

   • Entry points are high-level/human-readable
   • No god functions (1 function = 1 thing)
   • Comments are practical, not redundant
   • No orphaned code paths

4. UI/UX Flow Review

   • Onboarding flow (new user → first transcription)
   • Upgrade flow (free → pro)
   • Settings organization
   • Error states and recovery

---

Sub-Agent Strategy

Each of us will spawn:

• 1x Haiku agent - R&D / file scanning / grep work
• 1x Opus agent - Deep reasoning on findings, writing recommendations

Report findings back to main agents for consolidation.

---

Deliverables

1. Per-module findings - Issues, refactor opportunities, risks
2. Feature gate alignment matrix - Which gates exist where
3. Recommended fixes - Prioritized by launch-blocking vs nice-to-have
4. UI/UX flow audit - Screenshots/notes on any friction

---

Timeline

• Phase 1: Parallel deep dive (2-3 hours)
• Phase 2: Cross-reference findings (30 min)
• Phase 3: Consolidate into actionable PR(s) or issues

---

Open Questions for Dylan

1. Any specific flows that are known pain points?
2. Beta launch deadline (helps prioritize)?
3. Should we fix issues inline or just report them?