knowledgeintermediate

Documentation generator

A coding agent pointed at your source tree that keeps docs/ honest as the code changes, with a live markdown preview open next to the chat so you see the page update as it writes.

lines 34cost ~$0.30 per pass on a small repoapp id doc-writer

Install it locally

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

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

What you get

Docs that stay in sync with the code they describe
A live markdown preview pane, no separate tab to check
Atomic per-file edits, easy to review in the change log
The workspace pane config lives where it's actually read: ui.workspace

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: doc-writer5  name: "Doc writer"6  version: "1.0.0"7  category: "ai"89runtime:10  mode: conversation11  entry_agent: writer1213modules:14  filesystem:15    config:16      workspace: "."1718ui:19  workspace:20    render_mode: markdown21    entry_file: docs/index.md22    default_open: true2324agents:25  - id: writer26    modules: [filesystem]27    brain:28      provider: anthropic29      model: claude-sonnet-530      credential: { ref: anthropic_main, scope: per_user, provider: anthropic }31    system_prompt: |32      Read modules under src/. Update docs/ to match.33      One file at a time. Read before you write.34      Code blocks in docs must match the source exactly.

How it works

ui.workspace, not a fictional 'workspace module'

The live preview pane is UI configuration, declared under the top-level ui: block - ui.workspace.render_mode: markdown plus entry_file: docs/index.md. There's no such thing as modules.workspace.render_mode in the current schema.

filesystem is still what writes the files

The agent's actual read/write/edit access comes from the filesystem module as usual; ui.workspace only controls what the person watching sees, not what the agent can do.

system prompt enforces atomic edits

'One file at a time. Read before you write.' The agent reads a source module, updates the matching doc page, moves on - each edit lands as its own change you can review individually.

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 creativeLive React app builder/templates/react-builder knowledgeKnowledge base agent/templates/knowledge-base-agent opsScheduled data pipeline/templates/data-pipeline