# InDraft

Personal project — a config-driven LinkedIn drafting engine. Drafts in your voice on a cron, you review on your phone, nothing publishes without a passkey assertion.

- Status: live
- Role: Solo — engineering, deploy
- Period: May 2026 – Present
- Stack: Next.js 16, TypeScript, Vercel KV, Vercel Cron, OpenRouter, Claude Opus 4.7, Claude Haiku 4.5, WebAuthn (passkeys), Resend, Pexels, Tailwind v4
- Links: [Live](https://indraft.dev) · [Repo](https://github.com/feRpicoral/indraft)
- Canonical: https://picoral.me/projects/indraft

## What it does

You fork the repo, fill in `config.yml` with your profile paragraph, RSS feeds, posting schedule, and content pillars, and `yarn setup` deploys it on Vercel. From then on, a daily cron fires at 14:00 UTC, filters down to your configured days and local hour, pulls a fresh item from your sources, drafts a post in your voice, and emails you a magic link.

You open the review page on your phone. You can chat-edit the draft until you're happy with it, then tap "Post" — Face ID or Touch ID, and it's live on LinkedIn.

## Why I built it

LinkedIn post quality is a function of voice and timeliness. Most automation tools kill both: they generate generic AI prose and they post on a schedule that has nothing to do with what's actually happening that day. InDraft tries to preserve them — your voice is the model's primary grounding, the schedule is keyed off fresh sources, and a human is always in the loop before anything publishes.

It's also a generic, config-driven engine, not a SaaS. Nothing personal — voice, sources, credentials — lives in the repo. Fork it, point your config at your stuff, deploy.

## How it works

### Architecture

```
Vercel Cron (daily 14:00 UTC)
   | day/hour filter
Collector: RSS + GitHub
   |
Generator + anti-AI-tell linter
   |
Media: Pexels or owner upload
   |
Vercel KV
   |
Resend: magic-link email to /review
   |
Conversational chat edits
   |
WebAuthn passkey assertion
   |
LinkedIn Posts API
```

The state machine in `src/lib/state/drafts.ts` is the single writer of draft status and the load-bearing safety boundary. The only way to reach the LinkedIn Posts API is through a verified WebAuthn assertion — no scheduled or automatic path posts. An e2e test in `tests/e2e/publish-guard.test.ts` proves this at the state-machine level: a non-`PENDING_REVIEW` draft can never transition to `PUBLISHED`, and a captured assertion can't be replayed against an edited draft because the challenge is bound to the draft version.

### Key defaults

- **`link_placement: "none"`** — current LinkedIn-algorithm research shows external links suppress reach. InDraft surfaces the source URL inside the review UI instead. Override per post if you want.
- **Frontier model for drafting, cheap model for ranking** — Claude Opus 4.7 drafts because the anti-AI-tell bar degrades on weaker models; Claude Haiku 4.5 handles source ranking and mechanical tasks. Both routed through OpenRouter BYOK.
- **60-day LinkedIn token, no refresh** — self-serve LinkedIn apps issue 60-day tokens with no refresh path. InDraft stores `issued_at` in KV, emails a reauth ping ~7 days before expiry, and shows a "reconnect LinkedIn" banner in the review UI.

### Security model

- The only publish path is a click, then a WebAuthn assertion, then a server transition with a derived `publishProof`. The assertion challenge is `sha256(draft.id || draft.version || draft.body)`, so a captured assertion can't be replayed.
- Magic links are HMAC-signed, single-use (nonce stored in KV with a TTL), and expire after 24 hours by default.
- The public `/access` endpoint always emails the configured `NOTIFY_TO_ADDRESS` only — never an attacker-supplied target.
- Sessions are HttpOnly, Secure, SameSite=Lax cookies bound to a specific draft id.
- `.gitignore` excludes `config.yml`, `.env*`, and `.vercel/`. The setup script flows secrets `prompt -> vercel env add` only — never to disk.

## Status

Engine is feature-complete and MIT-licensed. The repo is the deliverable; my personal deployment lives at [indraft.dev](https://indraft.dev).

## Questions

### What is InDraft?

InDraft is an open-source LinkedIn drafting engine. A scheduled job pulls fresh items from your RSS feeds and GitHub activity, picks one, and drafts a post in your voice. You review on mobile, edit conversationally, and the post only goes live after a WebAuthn passkey assertion bound to that exact draft.

### Why a passkey for publishing?

There's no automatic publish path in the codebase. Every published post passes through a WebAuthn assertion bound to a sha256 of the draft id, version, and body — so a captured assertion can't be replayed against an edited draft. The state machine in src/lib/state/drafts.ts is the single writer of draft status and proves the property at the type level.

### How does InDraft preserve voice?

A free-text profile block in config.yml is the model's primary grounding. The drafting model is Claude Opus 4.7 via OpenRouter BYOK because the anti-AI-tell quality bar degrades on weaker models. A separate linter rejects em-dash spam, "let's dive in", and other tells before the draft hits the review UI.

### Why omit the link by default?

2026 LinkedIn algorithm research shows body-link posts get up to a 60% reach hit and the older "link in first comment" workaround gets up to 80%. InDraft's default is link_placement = "none" — the source URL is surfaced inside the review UI instead. You can override per post.