Lorebase
Editors publish articles; a workflow splits each one into passages and re-indexes it on every edit. Teammates ask in plain words: a server function retrieves passages with Postgres full-text search, re-ranks them with BM25 and has the app's AI answer from those passages only, with numbered citations. Without an AI key it answers extractively from the same passages. Thumbs-down answers and unanswered questions land on the editors' gaps board, where one click drafts the missing article.
Invite-only. Teammates ask and rate answers. Knowledge editors run the articles, passages and gaps.
What's inside
Ask page with cited answers, thumbs up or down and question history. Library of published articles with their indexed passages. Gaps board: unanswered and low-rated questions, draft-an-article in one click. Article, passage, source and citation management for editors. Weekly gaps digest by notification and email.
Tables
editorname, email, teammembername, email, departmentsourcename, kind, description, owner_iddocumenttitle, source_id, status, summary, body, url, version, word_count, …chunkdocument_id, source_id, doc_title, seq, heading, text, words, activequestionmember_id, text, answer, outcome, follow_up, mode, confidence, latency_ms, …citationquestion_id, chunk_id, document_id, rank, score, snippetfeedbackquestion_id, member_id, rating, comment
Roles
editorKnowledge editor, staffmemberTeammate, staff
Pages
/askAsk (member, editor)/libraryLibrary (member)/gapsGaps (editor)/documentsArticles (editor)/documents/:idArticle (editor, member)/sourcesSources (editor)/questionsQuestions (editor)/questions/:idQuestion (editor)/passagesPassages (editor)/citationsCitations (editor)/peopleTeammates (editor)/editorsEditors (editor)
Components
AskDeskAsk a question, read the cited answer, rate it, and revisit past questionsGapsBoardEditors' view of unanswered and low-rated questions, with draft-an-article and resolve actions
Rules on every write
feedbackYou can only rate answers to questions you asked
How its workflows run
Each one is a server workflow in your project: it runs on its trigger, with retries, and its runs show in the builder's log.
- Index a new article
When a row changes:
document created. Then: Look up chunk, then Update chunk, then Create chunk. - Re-index an edited article
When a row changes:
document updated. Then: Look up chunk, then Update chunk, then Create chunk, then Look up question, then Update question. - Answer a question from the knowledge base
When the app calls it:
server function (member, editor). Then: Look up member, then Create question, then Notify. - Draft an article from a question
When the app calls it:
server function (editor). Then: Look up question, then Create document, then Update question. - Flag answers marked not helpful
When a row changes:
feedback created. Then: Look up question, then Update question, then Notify. - Weekly knowledge gaps digest
On a schedule:
schedule weekly mon 09:00 over question. Then: Notify editor, then Send email.
Set it up
It starts a new project, with its data, roles and sign-in ready.
- Start the project
Use the button above, or ask your coding agent with the Appbricx MCP server connected (quickstart).
- Sign in as each role
The demo logins open each role's home. Change what you like in plain words or in the contract.
- Deploy
When dev looks right, deploy. Production gets a snapshot first, so you can roll back.
Use it with your coding agent
The agent creates the project, builds it and signs in as each role.
Start a new Appbricx project from the Lorebase template (slug "lorebase") with use_template. Call it "My Lorebase", then sign in as each role, verify it and tell me what to change first.{
"tool": "use_template",
"arguments": {
"slug": "lorebase",
"name": "My Lorebase"
}
}appbricx templates use lorebase --name "My Lorebase"Goes well with
Steward
Custom AI agent with tools, memory, a step-by-step run log and human approval before sensitive actions.
Sieve
Support triage agent: AI classifies every ticket with a confidence, routes it by your rules, drafts the reply; leads review the unsure ones.
Draftline
Multi-agent content pipeline: research, draft and edit agents hand off to each other, retry on failure, and a reviewer approves before publish.