Documentation

What the research portal is, how to choose a portal and find your way around.

Where the content comes from, what happens to each document, how a question is answered and what the portal checks before it shows you an answer.

Search answers by default - a short cited answer over the matching documents. Results only turns the answer off.

Ask questions conversationally, keep sessions, and run deep research.

How to read the confidence signals, citations and the evidence table.

Save a search or a question and be told when the corpus has something new for it.

The home surface - suggested questions, topic rows and a way in.

Browse, filter and page through the whole corpus.

The document view, its viewer, and asking questions of a single document.

A visual map of the corpus - an entity graph of what it is about, and a concept map of how its themes overlap.

Connect MCP clients to the portal's research through the knowledge box connector.

Turn a topic into a briefing, comparison, timeline, FAQ or pros and cons, with sources.

Build a knowledge check on any area of the corpus and test yourself.

Accumulate evidence around a research question over time.

Connect a knowledge box, then add, ingest and sync content into it.

Shape the topic taxonomy, the knowledge graph and per-document enrichments.

How the Help section works and how it stays separate from research.

How CorpusKit is built: tenants as data, retrieval behind one interface, provisioning as a product feature, and the API, front end and deployment pipeline.

Run a CorpusKit deployment: upgrades, credential encryption, the operator credential, portal lifecycle and limits, sign-in handoff, host aliases and search engines.

The credential model, working call shapes and architecture for building a research portal on Progress Agentic RAG, from the CorpusKit project.

What changed in each CorpusKit release, and what to check before upgrading.