Revenue channels 2026 — every product fans out to every viable channel via omni-publish

decision Sun Jun 21 2026 00:00:00 GMT+0000 (Coordinated Universal Time) decisionrevenuemonetisationdistributionomni-publishai-copynvidia-nimopenrouterautomationchannels

Revenue channels 2026 — fan-out via omni-publish

Decision summary

Every product the family ships — 26 apps + 17 packages + 5 books + future browser/VS Code extensions + CLIs + MCP servers — auto-publishes to as many revenue channels as 2026's API reality allows. Generated and orchestrated by @chirag127/omni-publish on every tag push (v*.*.*) in every repo. AI-generated platform-specific copy uses NVIDIA NIM (primary, free 5K req/mo) with OpenRouter free models as quota-failure fallback. Channels with dead 2026 APIs (Reddit OAuth closed, X free-tier no public writes, LinkedIn /v2 sunset, Medium tokens deprecated) drop into a single Telegram drafts channel with 4 sections, not auto-posted.

Channel matrix per surface type

Web app (subdomain on *.oriz.in)

npm package

Book

See full pipeline in book-publish-pipeline.md. Summary row here:

Extension (browser + VS Code)

Dead in 2026 — do NOT propose

Documented because earlier prompts kept suggesting these:

• Reddit OAuth — closed June 2026; the third-party API access tier was wound down after the 2023 API pricing war escalated to full closure for non-Reddit-app accounts.
• X API free tier — no public writes for free accounts since 2023; "Basic" tier is $100+/mo (paid ? fights no-card-on-file).
• LinkedIn /v2 — sunset; restricted to Marketing Developer Platform partners only.
• Medium integration tokens — deprecated 2023, still partly works but unreliable; ship-day surprises common.
• Amazon KDP API — never existed publicly; only Amazon Marketing Stream + KDP Publisher tools.
• Apple iBooks Author — replaced by Pages, Mac-only, no Linux path (per linux-ci-only rule).

These are listed so future prompts can recognise them as dead ends without re-grilling.

Manual-platform queue ? Telegram drafts channel

For X / Reddit / LinkedIn / Medium where automation is gone, omni-publish generates ready-to-paste drafts on every tag push and posts them to one Telegram channel (TELEGRAM_DRAFTS_BOT_TOKEN + TELEGRAM_DRAFTS_CHAT_ID) split into 4 sections by hashtag:

• #draft-x — 280-char teaser + URL + 2-3 hashtags
• #draft-reddit — title (=300 chars) + body (markdown) + suggested subreddit list
• #draft-linkedin — long-form post (3-paragraph, line-broken) + URL
• #draft-medium — full canonical article with <!-- canonical: <url> --> first-line comment

User copy-pastes each to its platform when convenient. No automation lies = no shadow-ban risk = no broken-link claims.

AI copy generation

Platform-specific copy (dev.to title vs Bluesky 300-char teaser vs LinkedIn 3-paragraph) is generated, not hand-written.

• Primary: NVIDIA NIM — 5,000 requests/month free, no card required, models include meta/llama-3.3-70b-instruct + nvidia/llama-3.1-nemotron-70b-instruct. Env: NVIDIA_NIM_API_KEY.
• Fallback: OpenRouter free models — meta-llama/llama-3.3-70b-instruct:free + deepseek/deepseek-r1:free. Env: OPENROUTER_API_KEY. Free tier ~50 requests/day per model; aggregate ~100/day across both.

omni-publish picks NIM by default; falls back to OpenRouter on HTTP 429 or 5xx. If both fail, the platform's row goes to the Telegram drafts channel with a [ai-fallback-needed] tag instead of being skipped silently.

Rate limit ceiling

Max 1 auto-post per channel per day per repo. Prevents:

• dev.to spam flag (their rule: =3 posts/24h per account; we stay well under)
• Hashnode flag (similar)
• Bluesky rate-limit (300 events/h account-wide)
• Mastodon instance soft-bans for new accounts

omni-publish keeps a JSON state file at ~/.config/oriz/omni-publish-state.json (CI: artifact across runs via cache) tracking last-post timestamp per (channel, repo). Tag pushes that violate the ceiling are deferred to next-day cron rerun.

Trigger

Every repo's .github/workflows/release.yml calls the omni-publish reusable workflow on tag push matching v*.*.*:

on:
  push:
    tags: ['v*.*.*']
jobs:
  cross-post:
    uses: chirag127/omni-publish-npm-pkg/.github/workflows/cross-post.yml@main
    with:
      title: "${{ github.event.head_commit.message }}"
      body_path: "RELEASE_NOTES.md"
      canonical_url: "https://github.com/${{ github.repository }}/releases/tag/${{ github.ref_name }}"
      type: "app"  # or package | book | extension
    secrets: inherit

Trigger payload

The type field determines which row-set from the matrix above fires. A package release never tries Play Store; a book release never tries npm.

The 5 first books

Status: planning (none have manuscripts yet). Locked names + scope in book-publish-pipeline.md:

1. Oriz Stack — Astro 6 + Cloudflare + Firebase Spark family architecture (technical, $19/$39)
2. Oriz Paisa: Credit Cards India 2026 — cards-app companion (Indian finance, ?499/?999)
3. Oriz PDF: From Browser to Native — PWA ? PWABuilder ? Play Store walkthrough ($14)
4. Oriz Janaushdhi: Generic Medicines India — janaushdhi-app companion (public health, ?299)
5. Oriz Me: 100-Year Strategy — personal essays from oriz-cs-me-app + 100-year-strategy doc ($9 PWYW)

Cross-refs

• The 17 packages this catalog covers ? [[architecture/the-23-packages]]
• The MIT relicense that unlocks free-for-OSS distribution perks ? [[decisions/ops/mit-license-all-repos]]
• The PWA ? native pipeline behind Play / Microsoft Store rows ? [[decisions/frontend/pwabuilder-as-primary-converter]]
• The blog-post fan-out engine ? [[decisions/content/cross-post-engine]]
• The npm package that orchestrates this ? [[decisions/packages/omni-publish-package]]
• The catalog hub at packages.oriz.in ? [[decisions/packages/packages-oriz-in-catalog]]
• Full book pipeline ? [[decisions/content/book-publish-pipeline]]