Delivery Channels

Agentic Universe publishes one curated digest per weekday. Pick whichever delivery channel fits your workflow — they all carry the same article set, just formatted differently.

Channels at a glance

Web dashboard

• / — today's top stories with scores, summaries, source attribution
• /archive — full paginated archive
• /best — all-time best (score ≥ 8)
• /articles/:id — deep link for a single article (shareable, cacheable)

Feeds

• RSS 2.0: /feed.xml — standard RSS with <category>, <pubDate>, <description> set to summaryShort
• JSON Feed 1.1: /feed.json — same 50 items in the modern JSON Feed spec

Both endpoints return the 50 most-recently-processed articles and are Cache-Control: public, max-age=3600 at the edge.

Telegram

• Broadcast channel (unlimited subscribers, best analytics)
• HTML parse mode — <b>, <i>, <a href>, inline keyboard buttons for "Read more"
• Delivered via grammY, a TypeScript-first bot framework
• Rate-limit-safe: one message per day, well under Telegram's 30 msg/sec broadcast ceiling
• Link the channel with a discussion group for per-article comments (optional)

Slack

• Modern Slack App using @slack/web-api + optional Bolt.js listeners
• Posts to a workspace-configured channel ID (SLACK_CHANNEL_ID)
• Message format: Block Kit — header block, section blocks with mrkdwn summaries, accessory "Read more" buttons
• Classic bots are deprecated; the app uses OAuth + bot-token auth

Email

• Personalized digest per subscriber (categories + keywords + score threshold)
• Double opt-in, unsubscribe at /api/unsubscribe?token=...
• One-click unsubscribe header (List-Unsubscribe) for inbox-provider compliance

Reliability

Delivery runs from the GitHub Actions pipeline after articles are processed and stored:

[7. Process Results] → [8. Deliver]
                         ├── Telegram (grammY)
                         ├── Slack (@slack/web-api)
                         ├── Email (personalized)
                         └── ISR webhook → Vercel revalidation

All downstream delivery calls run in parallel via Promise.allSettled. A single channel failing is logged but does not block the others — the pipeline still publishes to every channel that succeeded.

SLA

• Publish target: ~06:15 UTC on weekdays (primary cron 06:00 UTC, backup 09:00 UTC)
• If the 06:00 run fails: the 09:00 backup catches up; ISR revalidation fires on completion
• If all runs fail: the pipeline health monitor raises a Telegram alert to the operator — no empty digest is published

Notes for integrators

• The feeds + API follow the same rate-limit policy as the API reference: 100 req/min per source IP
• If you're building on top of the API or feeds, consume at most once per hour — the upstream pipeline only publishes daily