Understanding Context
Context is one of the most powerful features in Chibi AI, enabling you to create more relevant and accurate content. The perfect context combined with incredible AI prompts leads to the absolute best results possible from AI.
This guide will help you understand how context works and how to use it effectively across different parts of Chibi.
What is Context?
Context refers to all the information available to the AI model when generating content. Think of it like giving background information to a colleague before asking them to help with a task. The more relevant context they have, the better they can assist you.
In Chibi, context includes:
Your prompts and instructions
Content from your document
Information stored in memory fields
Global variables
Role instructions and settings
Why Context Matters
Understanding context in Chibi helps you:
Ensure the AI has the necessary information to generate appropriate content
Control token usage and costs
Maintain consistency in your content
Get more accurate and relevant results
Ultimately, to achieve the pinnacle of quality AI content
How Context Works in Different Parts of Chibi
Before we dive into context in Chibi, I want to mention this UI:
The highlighted part shows tools designed to help you pick from the various context variables you're about to learn. In Chibi, you will see this throughout. Use them to work faster and use context as you need.
Document Editor Context
When working in the document editor, Chibi provides several ways to access context through variables:
Document-Wide Variables
{{WholeDocument}}
: Captures everything in your document{{Title}}
: The document's title{{FirstParagraph}}
: The opening paragraph{{FirstSentence}}
: The first sentence of your document
Cursor-Based Variables
{{AboveCursor}}
: About 4,000 tokens of text above your cursor{{AfterCursor}}
: About 4,000 tokens of text after your cursor{{AfterCursorHalf}}
: About 1,000 tokens of text after your cursor{{LastParagraph}}
: The paragraph right before your cursor{{LastSentence}}
: The last sentence before your cursor{{ParagraphAfterCursor}}
: The paragraph immediately following your cursor
Selection-Based Variables
{{Selection}}
: Any text you've selected{{HighlightedWords}}
: Words you've highlighted using text formatting
Tool-Specific Context
Different tools in Chibi handle context in specific ways:
Inline Prompts (/prompt)
Uses your prompt
Includes active memory fields (memory stack)
Incorporates ~500 tokens of text above the prompt location
Can access any document variables
Write Button
Uses active memory fields
Includes ~500 tokens of text above the cursor
Follows the current role's instructions
Can access document variables
Finish Button
Uses a smaller context window (~125 tokens above cursor)
Does not include memory fields
Focuses on completing the current sentence naturally
Memory Context
{{MemoryStack}}
: Combines all active memory fieldsIndividual memory fields accessible by their labels
Memory fields can be document-specific or global
Context in Actions
Actions handle context through their task sequence, with context flowing from one task to the next. Each task can access:
Results from previous tasks using task variables
Document variables
Memory fields
Global variables
Note: For a detailed explanation of how context flows through actions, refer to our Action Context Flow guide.
Managing Context Effectively
Best Practices
Be Selective with Context
Include only relevant information
Use appropriate variables for your needs
Consider using smaller context windows for simple tasks
Use Memory Strategically
Store frequently needed information in memory fields
Use global memory for cross-document consistency
Activate only the memory fields you need
Optimize Token Usage
Choose cursor-based variables appropriate for your task
Use
{{AfterCursorHalf}}
instead of{{AfterCursor}}
when less context is neededConsider using Run Prompt instead of Stream Prompt when you don't need to write to the document immediately
Context Window Considerations
Different AI models have different context window sizes (the maximum amount of tokens they can process). When working with large amounts of context:
Consider chunking content for large documents
Use the Chunk Content task in actions when needed
Monitor token usage to optimize costs
Choose models with appropriate context windows for your needs
Advanced Context Tips
Context Cut-off
Use
\\\
in your document to create a context cut-off pointText above this marker won't be included in variables like
{{AboveCursor}}
Useful for controlling exactly what context is available
Role-Based Context
Roles can include variables in their instructions
This allows for dynamic context in your role definitions
Useful for creating more adaptive AI behaviors
Tool Integration
Various tools (Rework, Ask Cheebs, etc.) have their own context variables
These can be accessed in actions and prompts
Examples:
{{ReworkInput}}
,{{AskCheebsInput}}
,{{WebComposerInput}}
Conclusion
Understanding context in Chibi AI is key to creating effective content and workflows. By mastering the different ways context can be accessed and controlled, you can:
Create more accurate and relevant content
Build more efficient actions
Optimize your token usage
Maintain consistency across your documents
Remember that context is not something to avoid, but rather a powerful tool to harness. The key is finding the right balance of context for each specific task or prompt.
For more advanced information about how context flows through actions, check out our Action Context Flow guide.