Internal publishing system

Faceless Bot turns monitored news into approved Instagram posts.

The repo is scaffolded up to the Convex checkpoint: dashboard shell, typed data model, and setup contract are in place. The backend wiring resumes after local Convex onboarding.

Current internal snapshot

profiles

3

Instagram profiles staged

feeds

8

RSS sources mapped to profiles

drafts

12

Reviewable posts in queue

cadence

6/day

Default posting target

The next blocking step is Convex initialization. No public access, no unauthenticated app surface, and no publishing automation beyond the staged shell yet.
Source intake
Track RSS feeds per Instagram profile, pull article metadata, and prepare full-text extraction.
AI draft engine
Score relevance by topic, generate captions, and create a consistent image treatment through OpenRouter.
Approval queue
Review caption, image, and scheduled slot before the post is sent to Buffer.
Internal access
Google sign-in with company-domain allowlisting keeps the app private to your team.
What is already built

Interface

Login screen, dashboard routes, profile detail views, drafts table, article tracker, and settings pages are scaffolded and navigable.

Contract

Environment variables, data shapes, and the Convex setup checkpoint are documented so the remaining backend phases land on a stable contract.

Planned publishing loop

RSS feed pulled and normalized by profile.

Article scored for topic relevance and converted into a draft.

Caption and image generated in a consistent visual style for Instagram.

Reviewer edits if needed, approves, and Buffer receives the scheduled post.