knowledgebeginner

Knowledge base agent

Point this agent at a folder of docs and it answers questions the way a support engineer who actually read them would - citing the source instead of guessing. Built entirely on the rag module's real tools: ingest, query, and the rest.

lines 26cost ~$0.01 per answered questionapp id knowledge-base-agent

Install it locally

# save the YAML below as app.yaml in a new folder
digitorn install ./knowledge-base-agent

This is a local install: the folder you create is the source. No Hub package to publish first.

What you get

Answers grounded in your own documents, not the model's guesswork
A citation on every answer that actually used the knowledge base
One tool grant (rag.query) - the smallest template on this page
Room to grow: ingest, ingest_file and ingest_directory are one grant away

Credentials this template uses

Anthropic Claude

Click each one for the credential setup command, scope guidance, and provider details.

The full YAML, ready to fork

app.yaml

1schema_version: 223app:4  app_id: knowledge-base-agent5  name: "Knowledge base agent"6  version: "1.0.0"7  category: "ai"89modules:10  rag: {}1112runtime:13  mode: conversation14  entry_agent: helper1516agents:17  - id: helper18    modules: [{rag: [query]}]19    brain:20      provider: anthropic21      model: claude-haiku-4-522      credential: { ref: anthropic_main, scope: per_user, provider: anthropic }23    system_prompt: |24      Answer using the rag.query tool against the app's knowledge base.25      Always cite the source. If nothing relevant comes back, say so26      plainly instead of guessing.

How it works

modules.rag, not an invented voice/RAG hybrid

rag: {} with defaults is enough to get started - embedding model, chunking and citations all have sane production defaults. Add sources: [...] later to point it at a real folder.

one real tool: query

The helper agent is granted only {rag: [query]} - the actual tool name the module registers for asking a natural-language question against the knowledge base.

the system prompt sets the honesty bar

'If nothing relevant comes back, say so plainly instead of guessing.' A knowledge agent that quietly falls back to guessing is worse than one that admits it doesn't know.

Read the deep dive

10 apps you can ship in 50 lines of YAML

Read article

Newsletter

Get the next post in your inbox.

Engineering notes from the Digitorn team. No marketing, no launch announcements, no "10 prompts that will change your life". Just the things we write that we'd want to read.

Other templates

developerStarter coder/templates/starter-coder developerClaude Code clone/templates/claude-code-clone researchDeep research agent/templates/deep-research messagingDiscord helper/templates/discord-helper developerPull request reviewer/templates/pr-reviewer opsWeekly digest reporter/templates/weekly-digest knowledgeDocumentation generator/templates/doc-writer creativeLive React app builder/templates/react-builder opsScheduled data pipeline/templates/data-pipeline