opsadvanced

Scheduled data pipeline

Runs nightly, queries a read-only database connection with a DSN kept out of the YAML entirely, summarises the day's signal, and posts it to Slack. Demonstrates the real database module, a real cron provider, and pieces together.

lines 52cost ~$0.04 per runapp id data-pipeline

Install it locally

# save the YAML below as app.yaml in a new folder
digitorn install ./data-pipeline

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

What you get

Daily summary of yesterday's signal, delivered to Slack
A read-only database connection, capped at 500 rows per query
The DSN lives in an env var, never in app.yaml
Built on the real module id: database, not an invented 'postgres' one

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: data-pipeline5  name: "Scheduled data pipeline"6  version: "1.0.0"7  category: "devops"89runtime:10  mode: background11  entry_agent: pipeline1213tools:14  capabilities:15    default_policy: auto16    grant:17      - { module: pieces, tools: ["*"] }18  modules:19    database:20      config:21        databases:22          - name: prod_replica23            kind: postgres24            dsn_ref: PROD_REPLICA_DSN25            security:26              mode: read_only27              max_rows: 50028    pieces:29      constraints:30        allowed_pieces: [slack]31    channels:32      config:33        providers:34          nightly:35            adapter: cron36            config:37              schedule: "0 4 * * *"38            activation:39              agent: pipeline40              message: "Summarise yesterday's signal."4142agents:43  - id: pipeline44    modules: [{database: [query]}]45    brain:46      provider: anthropic47      model: claude-haiku-4-548      credential: { ref: anthropic_main, scope: per_user, provider: anthropic }49    system_prompt: |50      Query yesterday's user activity from prod_replica. Summarise top 351      signals. Post to the #data-summaries Slack channel via the pieces52      Slack action. Keep it under 250 words.

How it works

modules.database, with the real config shape

The module id is database (there's no separate 'postgres' module). Each entry under databases: names a connection, its engine (kind: postgres), and dsn_ref - the name of a server-side env var, so the connection string never touches the YAML.

security.mode: read_only, enforced

security.mode: read_only plus max_rows: 500 caps what a compromised prompt could ever do with this connection - even a successful injection can't write, and can't walk the whole table.

cron + pieces, the same real primitives as weekly-digest

The nightly schedule uses the same compiler-verified cron adapter shape, and the Slack post goes through pieces - no invented modules anywhere in this file.

Read the deep dive

How credentials work on Digitorn: a vault driven from 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 knowledgeKnowledge base agent/templates/knowledge-base-agent