Skip to main content

Overview

Workflow recording is ondoki’s core feature. It captures each user action — clicks, keystrokes, navigation — as an individual step with a screenshot, window title, and position data. After recording, AI auto-processes the workflow to generate titles, summaries, tags, and per-step descriptions.

Recording Methods

Both methods upload recordings to ondoki via the process recording API.

How Recording Works

1

Start a Recording Session

The desktop app or Chrome extension creates a new recording session via POST /api/v1/process-recording/session/create.
2

Capture Steps

Each user action creates a step with:
  • Screenshot of the current screen/window
  • Action type (click, keystroke, scroll, navigation)
  • Click position (global and relative coordinates)
  • Window title and size
  • Timestamp
  • Key pressed or text typed (for keyboard actions)
3

Upload to ondoki

Steps and screenshots are uploaded to the server. Metadata is sent as JSON, screenshots as image files.
4

Finalize the Workflow

The client calls the finalize endpoint, which triggers AI auto-processing.

AI Auto-Processing

When a workflow is finalized (and an LLM is configured), ondoki automatically generates:
Generated FieldDescription
TitleDescriptive title based on the recorded steps
Summary50–150 word overview of the workflow
TagsTopic tags for categorization
DifficultyEasy, Medium, or Advanced rating
Time EstimateEstimated completion time
Step AnnotationsPer-step titles and descriptions explaining what happened
Guide MarkdownComplete step-by-step guide in Markdown format
AI processing requires an LLM to be configured in Project Settings → AI/LLM. Without it, workflows are saved but not auto-annotated.

Workflow Editor

After recording, you can edit any workflow in the browser:
  • Reorder steps — drag-and-drop or use the reorder API
  • Edit step descriptions — refine auto-generated annotations
  • Delete steps — remove unnecessary or duplicate steps
  • Add steps manually — insert additional steps with descriptions
  • Re-annotate — trigger AI annotation on individual steps via the annotate button
  • Set metadata — change title, assign to folder, set privacy, add icon

Organizing Workflows

Workflows support the same organizational features as documents:
  • Folders — nested folder hierarchy with drag-and-drop
  • Projects — team-scoped containers
  • Privacy — mark workflows as private (owner-only) or shared
  • Icons — assign Tabler icons or favicons with custom colors

Exporting Workflows

Workflows can be exported in multiple formats:
FormatHow
PDFGET /api/v1/process-recording/{id}/export?format=pdf — uses Gotenberg
MarkdownGET /api/v1/process-recording/{id}/export?format=markdown
Public LinkGenerate a share token for read-only access without login
GitExport project workflows to a Git repository via Git Sync

Soft Delete and Recovery

Deleted workflows are soft-deleted (moved to trash) and can be restored:
  • Trash view: See all deleted workflows in a project
  • Restore: Recover a soft-deleted workflow
  • Permanent delete: Permanently remove a workflow and all its data