System Prompts

System Prompts

System Prompts are reusable instruction presets for workspace chat behavior.

Open the preset management page from Settings > Agents > System Prompts.

Use them when you want instruction templates that can be reused across multiple workspaces instead of rewriting each workspace prompt from scratch.

What system prompts affect

System prompts shape how a workspace responds before the user prompt is processed.

They are a good place for:

  • tone and style rules
  • domain constraints
  • formatting requirements
  • response policies
  • reusable organizational guidance

They are not the same thing as one-off prompt commands or file-based personality data.

Where presets are managed vs applied

RealTimeX splits this workflow into two surfaces:

  • Settings > Agents > System Prompts: create and manage reusable presets
  • Workspace Settings > Chat: apply or edit the system prompt used by a specific workspace

This means the preset library is shared, but the final active prompt still belongs to each workspace.

Managing the preset library

The System Prompts management page supports:

  • search
  • filtering between All, My created, and Purchased
  • create
  • edit
  • delete
  • copy prompt text
  • publish or update publish to Marketplace

Each preset has:

  • name
  • optional description
  • prompt body

Creating a preset

  1. Open Settings > Agents > System Prompts.
  2. Click Create.
  3. Add a name.
  4. Add an optional description.
  5. Write the prompt body.
  6. Save it.

After saving, the preset becomes available from the workspace prompt editor.

Applying a preset to a workspace

Go to:

  • Workspace Settings > Chat

In the System Prompt area, current builds support:

  • direct editing
  • Quick Apply Preset
  • prompt history
  • publishing the current prompt to Marketplace

Quick Apply Preset is the fastest way to seed a workspace with a managed system prompt template.

Variables inside system prompts

System prompts can reference managed variables such as:

  • {date}
  • {time}
  • {datetime}
  • {user.name}

The workspace editor links directly to the variable manager and shows available placeholders.

For the full variable guide, see System Prompt Variables.

Good uses

System prompt presets work best for reusable baseline behavior such as:

  • "answer as a finance operations assistant"
  • "format every answer as a structured checklist"
  • "respond in a given language unless the user asks otherwise"
  • "follow a specific support policy or escalation style"

System prompts vs slash commands

Use System Prompts when:

  • the behavior should apply automatically
  • the instruction is part of the workspace's baseline operating rules
  • you want consistent behavior across many chats in the same workspace

Use Slash Commands when:

  • the user should invoke the behavior manually
  • the instruction is situational rather than always-on

System prompts vs personality

Use System Prompts when:

  • you need a single active instruction block for chat behavior
  • the behavior is relatively compact and workspace-specific

Use Personality when:

  • you want a broader file-based instruction or identity layer
  • the behavior needs richer structure than one prompt field
  • the configuration should live as durable files rather than one text area

Marketplace behavior

System prompts can also be local, purchased, or published.

That makes them useful for:

  • rolling out common prompt strategies across teams
  • reusing approved prompt baselines across workspaces
  • distributing curated prompt packs through Marketplace

Troubleshooting

I created a preset but do not see it in a workspace

Check:

  • that the preset saved successfully
  • that you are editing Workspace Settings > Chat
  • whether the preset is being filtered or not loaded yet

The workspace behavior still does not match the preset

Check:

  • whether the workspace prompt was actually replaced or edited afterward
  • whether System Prompt Variables contain the values you expect
  • whether other workspace configuration is also shaping the output

Related guides