HOARDCORE

Open source · self-hosted · early access

Track the deal.
Keep the evidence.

Catalogs change. Prices move. Stock disappears. Hoardcore turns scattered inventory into a searchable history, then keeps the research behind your next buying decision in one place.

Explore the code Read the deployment guide

Run it on your own infrastructure. Your catalog observations and research stay in your PostgreSQL database.

Source-neutral coreObservation historyProvider-neutral researchOperator-controlled collection

What works today

From source signal to usable context.

Hoardcore is already a working tracker, not just a schema or a landing page. The first source module supports Shopify storefronts and collections; one installation can register many independent sources.

01

Collect conservatively

Run bounded, paced catalog collections with access-policy checks and visible run logs. Configure manual or per-source schedules; pause a source whenever you need to.

02

Review what changed

Browse current listings with filters, saved views, stock state, pricing, images, and source evidence. An overview surfaces inventory totals and changes across sources.

03

Keep research attached

Export versioned research packets, bring structured results back with validation and preview, or add market comparables manually. A scoped API supports coding agents.

04

Watch what matters

Watch individual listings or save filter views. Opt in to ntfy notifications for new matching observations without turning every source change into noise.

The workflow

Notice it.
Investigate it.
Remember why.

Hoardcore separates facts collected from sources from the research and judgment you add later. A current listing never has to erase the trail that led you there.

See the module model
  1. 01

    Register sources

    Add supported storefronts or collection scopes independently.

  2. 02

    Collect observations

    Choose when to run, watch progress, and retain source evidence.

  3. 03

    Review & research

    Filter inventory, inspect changes, and attach market context.

  4. 04

    Act on a signal

    Watch a listing, receive opt-in alerts, or keep investigating.

Small on purpose

Your stack. Your data.

One application container, one PostgreSQL database. The durable job worker runs inside the app process; committed migrations run there on startup too. There is no required Hoardcore account service, separate worker, migration sidecar, or object store.

Sign in through an operator-chosen OIDC provider. Existing linked accounts can optionally use email magic links when SMTP is configured.

Deployment details

Still growing

Built in the open.

This is early software. More source modules, complete-snapshot presence tracking, and deterministic opportunity calculations are next. We won't claim an absent listing is “gone” based on an incomplete collection.

Follow development Read project status