SmellyFeet — Information Broker Frontend

A server-rendered Go web frontend for Information-Broker. Browse AI-generated summaries of cybersecurity articles, read the full stored text, and jump to the original source. Supports search, feed filtering, pagination, and a stats page.

About

SmellyFeet is the reading room for Information Broker — a continuously-running intelligence pipeline that scrapes cybersecurity RSS sources, summarizes every article with a local LLM, and stores them for fast browsing.

This app turns that firehose into something you can actually read: a clean stream of AI-written summaries, a full-text view for each article with a link straight back to the original source, plus keyword search, per-source filtering, and live system statistics.

RSS sources → LLM summarize → PostgreSQL → Information Broker API → SmellyFeet

It's a small, dark-themed Go application with no database of its own — it talks only to the Information Broker HTTP API, so it deploys and restarts independently of the scraper. An in-app version of this page is served at /about.

Run

cp .env.example .env   # edit API_BASE_URL / PORT as needed
go run .

Then open http://localhost:3000.

Styling is a pre-built, committed CSS file. If you change templates or the theme, regenerate it with ./scripts/build-css.sh (downloads the Tailwind standalone CLI on first run).

Configuration

Variable	Default	Description
API_BASE_URL	http://localhost:8080	Base URL of the Information-Broker API
PORT	3000	Port the frontend listens on

Public deployment

The site is designed to be exposed through a Cloudflare Tunnel — see deploy/README.md. The app sends per-route Cache-Control headers (articles cache at the edge for an hour), strict security headers, and serves everything — CSS, fonts, favicon — from the binary with zero JavaScript and zero third-party requests. An Atom feed is available at /feed.xml.

Meetups

A community-run directory of infosec meetups, in the BSides spirit. Listings live in an embedded seed file rather than a database, so the app stays a stateless, GET-first proxy.

Routes

Route	Description
GET /meetups	List upcoming/past meetups. Filters: ?city=, ?tag=, ?online=1, ?chapter=
GET /meetups/{slug}	Meetup detail page
GET /meetups/{slug}/ics	Download an .ics calendar file for the meetup
GET /meetups/chapters	Chapter directory (discovery only — see attribution below)
GET /api/meetups	Read-only JSON of the current listings. Filters: ?city=, ?tag=, ?from=, ?to= (RFC3339)

All of the above (and the nav tab) are gated by MEETUPS_ENABLED.

Configuration

Variable	Default	Description
MEETUPS_ENABLED	true	Gates the Meetups tab and every /meetups* + /api/meetups route
MEETUPS_DEFAULT_TZ	Europe/London	IANA timezone used to display meetup times

Adding a meetup

There's no database — meetups are published by editing the embedded seed, internal/server/meetups_seed.json:

1. Add an entry with: slug, title, summary, description, starts_at/ends_at (RFC3339 with a UTC offset, e.g. 2026-09-24T18:30:00+01:00), location_type (in_person/online/hybrid), venue_name/venue_address, city/region/country, online_url, rsvp_url, chapter_name, chapter_url, and tags (a string array).
2. Every URL field (online_url, rsvp_url, chapter_url, and a chapter's website) must be http:// or https:// — anything else fails validation at startup.
3. If you also changed templates, regenerate CSS with bash scripts/build-css.sh.
4. Commit and redeploy. Publishing is just "edit the seed, redeploy" — git history is the audit log.

Refreshing tracked events

The tracker refreshes fully autonomously: a weekly cron on the host runs scripts/cron-refresh.sh, which resets to origin/main, re-scrapes every source with scripts/import-events.py --apply, and rebuilds the container. No human in the loop and no hand-pinned events — the tracker is exactly what the sources yield each week. A safety floor skips the deploy if a scrape collapses (< 30 events). Install the cron with:

crontab -l 2>/dev/null | grep -q cron-refresh || \
  (crontab -l 2>/dev/null; echo "0 4 * * 1 /home/pure/smellyfeet-frontend/scripts/cron-refresh.sh") | crontab -

To refresh manually / preview without deploying:

Meetup events are tracked from each chapter's own website. To refresh them:

python3 scripts/import-events.py > /tmp/events.json

This pulls from BSides chapter sites, the cryptax/confsec conference list, the DEF CON forum calendar (incl. regional editions like DEF CON Middle East), and a few named conference sites, and prints candidate events. It is best-effort: heterogeneous sites mean a few dates may be CFP/ticket dates, and JS-only or cert-broken sites yield nothing (reported on stderr). Review the dates, then merge the good ones into the meetups array of internal/server/meetups_seed.json and redeploy. Manually-verified events (e.g. chapters whose date is only rendered client-side) can be pinned by hand.

Attribution

This is a community meetup tracker in the BSides spirit — it is not an official BSides mirror or endorsed listing. The official, approved chapter list lives at bsides.org/chapters, and every Meetups page links back to it.

Architecture

Talks to the Information-Broker HTTP API only (no direct DB). internal/apiclient wraps the API, internal/server renders html/template pages styled with Tailwind (CDN). See docs/superpowers/specs/2026-06-05-information-broker-frontend-design.md.

Test

go test ./...