Prompts

Create reusable AI prompt templates with dynamic variables, automatic version history, and built-in AI chat — all from your Context Repo dashboard.

Prompts are the core building blocks of your Context Repo workspace. They're reusable templates for AI interactions that you can create, organize, version, and access from any connected AI assistant through the MCP server.

You can add dynamic variables, track every change with automatic versioning, and even chat with an AI about your prompt content — all from the prompt detail page.

Creating a Prompt

Click New Prompt in the dashboard toolbar to open the creation dialog.

The dialog includes these fields:

Field	Required	Description
Title	Yes	A short, descriptive name for your prompt — this is how you'll find it in search and your prompt list
Description	No	A brief note about what this prompt does, shown in search results and the prompt list
Prompt Content	Yes	The full prompt text, optionally with dynamic `${variables}`
Collections	No	Assign your prompt to one or more collections to keep things organized

Once you click Create Prompt, you're taken directly to the prompt detail page where you can preview, edit, and start chatting with AI about your content.

Every prompt gets version history from the moment it's created. Your first save is automatically recorded as version 0.

Editing a Prompt

Open any prompt from your list to see its detail page. The detail page has four tabs: Preview, Editor, Versioning, and AI Chat.

To edit your prompt:

Switch to the Editor tab
Click the Edit button in the top-right corner
Update any field — title, description, content, or collection assignments
Click Save Changes when you're done

Your edits are saved as a new version automatically (see Version History below). If you change your mind, click Cancel to discard your changes.

The Preview tab shows your prompt content rendered as formatted text, along with metadata like creation date, last updated, visibility status, target model, and version number. If your prompt includes variables, they're listed below the content with their types.

Variables

Variables make your prompts dynamic. Instead of writing fixed text, you can add placeholders that get filled in when the prompt is used.

Syntax

Use ${variableName} anywhere in your prompt content to create a variable:

Review this ${language} code for ${reviewType} issues:

${codeSnippet}

When you or an AI assistant uses this prompt, each variable becomes a fill-in-the-blank field.

Variable Types

Each variable has a type that tells Context Repo what kind of value to expect:

Type	Description	Example
String	Free-form text input	A language name, a file path, or a paragraph of text
Number	Numeric values	Temperature settings, line counts, or thresholds
Boolean	True or false	Flags like "include examples" or "verbose output"

Default Values and Descriptions

You can set a default value for any variable, so it's pre-filled when someone uses the prompt. You can also add a description to explain what the variable is for — helpful when sharing prompts with others or revisiting them later.

Variables are detected automatically from your prompt content. When you include ${variableName} syntax, Context Repo parses and tracks them with their type, default value, and description.

Version History

Every time you save changes to a prompt, Context Repo automatically creates a new version. You don't need to do anything special — versioning happens behind the scenes.

What's Tracked

Each version records:

The full prompt content at that point in time
A change log describing what was updated
A timestamp and who made the change
A diff showing exactly what lines were added or removed compared to the previous version

Viewing History

Switch to the Versioning tab on any prompt detail page. You'll see a timeline of every version, with the current version marked at the top. Click the expand arrow on any version to view a line-by-line diff highlighting additions and deletions.

Restoring a Previous Version

If you need to roll back, click the Restore button on any previous version. You'll see a confirmation dialog before anything changes. Restoring doesn't delete history — it creates a new version with the content from the version you selected, so you never lose anything.

Managing Prompts

Visibility

Every prompt is either public or private:

Private prompts are visible only to you
Public prompts can be discovered by other Context Repo users

Toggle visibility from the actions menu (the three-dot icon) on the prompt detail page. Select Make public or Make private to switch.

Archiving

If you're not actively using a prompt but don't want to delete it, archive it. Archived prompts are hidden from your main list but can be restored anytime.

Archive: Open the actions menu and select Archive
Unarchive: Find the archived prompt and select Unarchive from its actions menu

Deleting

To permanently remove a prompt, open the actions menu and select Delete. You'll see a confirmation dialog — this action can't be undone.

Deleting a prompt removes it and all its version history permanently. If you might need it later, consider archiving instead.

Organizing with Collections

You can assign prompts to collections during creation or while editing. Collections help you group related prompts and documents together — for example, a "Code Review" collection might contain your review checklist prompt alongside relevant reference documents.

To update collection assignments, switch to the Editor tab, click Edit, and use the collections selector to add or remove collections.

AI Chat

Each prompt has a built-in AI chat interface. Switch to the AI Chat tab to start a conversation about your prompt content.

Starting a Chat

Click New Chat to create a chat session. The AI has access to your prompt content and can help you refine it, explain sections, suggest improvements, or answer questions about it.

Chat Sessions

Your conversations are organized into sessions. The sidebar on the left shows all your sessions for this prompt — click any session to resume where you left off. You can collapse the sidebar for a wider chat view.

Model Selection

Pick the AI model you'd like to chat with using the model selector in the chat interface. Available models are all from OpenAI, including GPT-OSS 20B, GPT-OSS 120B, GPT-4o, and GPT-5.1 Thinking — each with different strengths for speed and reasoning depth. Your last-used model is remembered across sessions.