Video Tutorials

Step 1: Enter prompt list

Enter one prompt. For a batch, enable Multi-Prompt and separate prompts with a blank line; you can also import a .txt file.

Step 2: Choose settings

• Model: Choose an available model for the selected provider and image/video mode.
• Aspect Ratio: The available ratios are loaded for the selected provider, model, and media type.
• Results per prompt: Set the quantity when the selected provider and model support it.
• Reference Images: Choose images in the reference picker. The allowed number depends on the provider, model, and video input mode.

Step 3: Click Generate

The extension takes over and:

• Submits prompts in sequence, or in parallel where the provider and run mode support it
• Shows provider errors and retries when the active execution path supports recovery
• Auto-downloads results if Auto Download is on
• Shows progress in the batch list and execution tracker

Common node types

How to Use

1. 1Open the Workflow tab and click Create New
2. 2Drag node tiles from the left toolbar onto the canvas
3. 3Drag from a node's output port to the next node's input
4. 4Click a node to fill in prompts, model, ratio, and refs
5. 5Click Run, review the confirmation, and track each node in Queue

Setup in 3 steps

1. 1Extension → Settings → MCP tab → Create new token (copy it — shown only once).
2. 2Copy the MCP server address and token from Settings → MCP into your AI client. The sample below is for Claude Desktop; other clients use their own setup screens.
3. 3Restart the AI client, check that TobyFlow tools appear, then ask it to check provider status before generating. Keep the selected provider signed in.

{
  "mcpServers": {
    "tobyflow": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://tobyflow.mcp", "--header", "Authorization:Bearer <token>"]
    }
  }
}

What the AI can do

• gen_image / gen_video — generate images & videos (voiceover on supported models)
• create_workflow / update_workflow / run_workflow — build & run visual workflows
• upload_ref — upload an image to the current Flow project; it does not attach the image to a generation automatically
• create_project / open_project / get_current_project — manage Flow projects
• get_provider_status / list_capabilities / list_voices — check readiness, models, limits & voices
• list_saved_results / add_saved_results_to_refs — inspect local batch results and add selected images to Gen references

Generation tools return a job_id. Ask your agent to poll get_job until it finishes and compare the requested result count with the returned results before reusing them.

Reference images

Adding saved results to Gen references only changes the Gen tab. For a later MCP generation, pass verified selectors as saved_refs or saved_refs_per_prompt; keep the same browser runner because files are stored locally.

Optional: auto-save MCP-generated files to disk (folder & resolution are configurable in the MCP tab).

Create an Album

In the Photos tab, create an Album and add images by uploading, dragging files in, or capturing an existing Flow result. Images remain available on this device across reloads and project switches.

@Mention in Prompts

Name each image with a slug (e.g. main_character), then reference it from any prompt:

Choose @Mention as the reference mode, select the image, and use its name in the prompt. For Flow, Toby Flow uploads a local image to the active project when needed; other providers use their own reference path.

Link Account

1. 1Open the Tools tab in the extension and click Telegram
2. 2Click Connect Telegram to get a one-time OTP code
3. 3In Telegram, open @tobyflow_bot and send /link [OTP]

Bot Commands (15 total)

Plus /start, /link, /unlink, /plan, /upgrade, /contact, /help

What you control

Preset Examples

• Product Photography: Hero / detail / lifestyle angles for e-commerce
• Character Turnaround: Front / three-quarter / side / back for character sheets
• Architecture: Exterior, eye-level, and aerial views for buildings

How to Use

1. 1Open the Tools tab, click Effects
2. 2Pick a source image (upload, album, or recent)
3. 3Choose a category and an effect; adjust intensity 0–100%
4. 4Click Generate — the result is submitted to Flow