Mechanically
Has never worked. Messages.tsx:22 reads process.env.N8N_WEBHOOK_URL in a Vite app where process.env does not exist, so the URL is always '' and every send POSTs to the empty string.
Broken
Domain 16 · platform · ~1,100 LOC
Messaging
Has never worked. Messages.tsx:22 reads process.env.N8N_WEBHOOK_URL in a Vite app where process.env does not exist, so the URL is always '' and every send POSTs to the empty string.
What it is and how it feeds the rest
How it connects
Nothing.
Worth knowing
UNRESOLVED CONTRADICTION between lanes. The domain and block lanes call this dead vestigial code; the competitor lane calls the missing conversation layer the single biggest gap versus the market. The research question is a decision, not a survey.
Pages
src/pages/Messages.tsx
Components
src/components/messages/
Convex modules
- none — no backend
Tables
- none
Findings that touch this domain
medium process.env in a Vite app — webhook URL is always empty string
src/pages/Messages.tsx:22
The decision — what we do here
BUILD (the fork) — never worked; the missing market half. Wrap chatwoot schema / zavu-inbox. Strategic: needs your fork decision.
Repos we analyzed for this domain
STUDY chatwoot/chatwoot
Fan-inbox schema — study the conversation/assignment model, don’t deploy the monolith.
LIFT zavudev/zavu-inbox
Rebuildable-mirror inbox boundary (TS) — the port-able piece.
LIFT melgarafael/DeskcommCRM
AI→human conversation state machine (claim/transfer/draft/handoff) in TS.
LIFT ArnasDon/wacrm
Shared-inbox + idempotent-webhook + handoff patterns (TS).
STUDY devennn/senqo
Manual-takeover / AI-mode boundary pattern.
WRAP block/buzz
Collaboration/agent plane (Camron chose it). Team coordination + fan-adjacent, NOT money.
LIFT = copy the code into the fork · WRAP = run underneath as a service · STUDY = architecture only. Verdicts from the OFM sourcing campaign.
Open questions
Research priority for this domain is highest. Agent findings splice in below this line and survive regeneration.