End-to-end demo

This page describes the intended beta demo path. A screenshot-based product tour will be added after the console screens and Docker beta quickstart are finalized.

Scenario

An infrastructure team searches internal documents for:

k8s pg timeout

Relevant documents use mixed terminology:

k8s, kube, kubernetes
pg, postgres, postgresql
timeout, deadline exceeded, connection timeout

SkeinRank helps the team govern those aliases and verify that runtime search uses the correct canonical terminology.

Demo walkthrough

1. Create a terminology profile

Create a profile named infra_incidents. Add canonical terms such as:

Canonical value	Slot	Aliases
kubernetes	TOOL	k8s, kube
postgresql	DATABASE	pg, postgres, psql
timeout	SYMPTOM	deadline exceeded, connection timeout

2. Review suggestions and guardrails

Use Suggestions to propose aliases without giving every contributor direct write access. Use Guardrails to block noisy values such as short or ambiguous aliases.

3. Create an Elasticsearch binding

Create a binding that applies infra_incidents to the documents index.

profile: infra_incidents
index: company_docs
text fields: title, body
target field: skeinrank.attributes
discriminator: team=infra

4. Run a dry-run

Dry-run previews what SkeinRank would write without changing Elasticsearch. This is useful before approving a dictionary change or running a full enrichment job.

5. Run enrichment

Run an enrichment job for the binding. The job applies the selected profile to documents in the selected index scope and updates runtime state.

6. Audit the runtime snapshot

Open Snapshots and confirm that the binding has an active snapshot. This answers the question:

Which immutable terminology version is runtime search using right now?

7. Verify Search Playground

Open Search Playground, select the binding, and run:

k8s pg timeout

Expected result:

canonical query: kubernetes postgresql timeout
matched aliases: k8s → kubernetes, pg → postgresql
runtime snapshot: infra_incidents@...

8. Integrate through API

After the behavior is verified, an application backend can follow the API integration guide and call SkeinRank by binding:

{
  "binding_id": 42,
  "query": "k8s pg timeout",
  "size": 10
}

The application can then combine SkeinRank output with its existing search or RAG pipeline. For an “All docs” scope, call /v1/search/multi with explicit binding IDs rather than guessing one global profile.

What this demo should prove

• SkeinRank is not just a dictionary CRUD UI.
• Profiles, bindings, and snapshots create a safe runtime model.
• Enrichment can be operated and audited.
• Search Playground shows the value immediately.
• External applications can integrate through binding-aware runtime APIs.