Skip to main content
The Writing Assistant refines one Post body item in the web editor. You can manage its Workspace settings in the web app, public API, or CLI.

Refine the copy

  1. Open a Post in the web editor.
  2. Click the sparkles icon below the body field.
  3. Select a refinement option.
OpenPMM shows a comparison between the source copy and the refined copy. Select Accept to use the refined copy. Select Reject to keep the source copy. The Workspace starts with these refinement options:
  • Fix grammar
  • Improve writing
  • Make it punchier
  • Condense
  • Rephrase / Mix it up
  • Expand & elaborate
  • Improve structure & spacing
The built-in options keep facts, links, mentions, hashtags, and the source language. They also remove common machine-written patterns. A Workspace member can change the title, prompt, and icon for each option.

Use a custom prompt

Type an instruction in the search field. Press Enter, or select Use custom prompt. OpenPMM applies the custom instruction and the channel rules. It does not apply the built-in humanization rules or the Workspace option settings.

Use channel-aware copy

The Writing Assistant uses the rules for the selected channel.
  • X copy uses short sentences and a strong opening.
  • LinkedIn copy uses useful opening lines and short paragraphs.
  • Threads copy uses a direct and conversational style.
  • Bluesky copy stays concise and respects its text limit.
  • Mastodon copy uses the limit from the selected destination.
  • Other channels use their native body format and text limit.
The Writing Assistant refines only the selected body item. It does not create new thread items or change a YouTube title.

Refine synced copy

A synced follower channel cannot use the Writing Assistant. Unsync that channel before you refine its copy. You can refine the master channel. OpenPMM sends the exact result to each compatible synced channel through the normal sync rules. It does not create a separate channel-specific result for a synced follower.

Customize Workspace options

The AI Refine settings apply to all members of the Workspace.
  1. Click the cog in the Writing Assistant menu.
  2. Change the Title field to change the menu label.
  3. Change the Prompt template field to change the refinement action.
  4. Click the icon button to open the icon picker.
  5. Select one of the 30 icons.
  6. Click Save.
You can change all seven options. For example, you can change Fix grammar to Translate to German. The prompt must tell OpenPMM to translate the copy. Click Reset for an option to restore its default title, prompt, and icon. Click Save to save the reset. Changing a setting does not change existing body text. The setting applies the next time a Workspace member selects the option.

Manage options through the API or CLI

Use GET /v1/workspaces/{workspace_id}/writing-refinement-settings to read the effective settings. The response always includes all seven stable action keys, including default values that the Workspace has not changed. Use PUT /v1/workspaces/{workspace_id}/writing-refinement-settings to replace the complete settings. Send the ETag from the GET response in If-Match. The request must include all seven actions. Each action requires a title, prompt, and supported icon key. The CLI performs the same workflow:
The update command reads the current ETag automatically. Pass --etag when a script must update only the exact version that it read earlier.

Review the system prompt

The settings dialog includes a System prompt information icon. Point to the icon or move keyboard focus to it to read the system prompt. The system prompt shows the source rules, channel guidance, content limit, and humanization rules. It also shows where OpenPMM adds your prompt template. Click the information icon to keep the system prompt open on a touch device. You cannot change the source rules, output format, or channel limits. A custom prompt does not use the humanization rules.