Skip to content

BuildOn Decorator

A meta-decorator that builds upon previous context or responses rather than starting from scratch. This enables continuity across interactions, allowing refinement, extension, or alteration of previous outputs in a coherent manner.

Category: Meta

Parameters

Parameter Type Description Default
reference enum What to build upon from the previous context last
approach enum How to build upon the referenced content extend
preserveStructure boolean Whether to maintain the structure of the referenced content True

Reference Options

  • last: Build specifically upon the most recent response or context provided.
  • specific: Build upon the specific part of the previous context mentioned in the current prompt.
  • all: Consider all previous exchanges in the conversation when building your response.

Approach Options

  • extend: Add new information, examples, or dimensions that extend and complement the existing content.
  • refine: Improve, correct, or enhance the existing content while preserving its core message.
  • contrast: Provide alternative perspectives or approaches that contrast with those in the existing content.
  • synthesize: Combine and integrate ideas from the existing content into a new cohesive whole.

Examples

Basic extension of the previous response

+++BuildOn
Add more detail about implementation challenges.

Extends the previous response by adding more detailed information about implementation challenges while maintaining continuity

Specific refinement with structural changes

+++BuildOn(reference=specific, approach=refine, preserveStructure=false)
Improve the section on risk analysis with more quantitative measures.

Refines specifically the risk analysis section from the previous content with more quantitative measures, potentially restructuring it

Model-Specific Implementations

gpt-4-turbo

Instruction: Continue from {reference} response, building on what was already said. {approach} the existing content while {preserveStructure}. Make sure your response feels like a natural continuation rather than starting over.

Notes: This model sometimes needs explicit reminders to fully incorporate previous context

gpt-4o

Instruction: Treating the {reference} as your foundation, please {approach} that content. {preserveStructure} Ensure your response maintains continuity with what came before.

Notes: This model handles contextual building well but benefits from clear instructions about which aspects to preserve

Implementation Guidance

Basic extension of previous response

Original Prompt: 

Add more detail about implementation challenges.

Transformed Prompt: 

Please build upon the previous context or response rather than starting from scratch, maintaining continuity across the interaction. Build specifically upon the most recent response or context provided. Add new information, examples, or dimensions that extend and complement the existing content. Maintain the same organizational structure, format, and overall framework as the referenced content.

Add more detail about implementation challenges.

Specific refinement with structural changes

Original Prompt: 

Improve the section on risk analysis with more quantitative measures.

Transformed Prompt: 

Please build upon the previous context or response rather than starting from scratch, maintaining continuity across the interaction. Build upon the specific part of the previous context mentioned in the current prompt. Improve, correct, or enhance the existing content while preserving its core message. Feel free to reorganize or restructure the content in a new way that better serves the current purpose.

Improve the section on risk analysis with more quantitative measures.

Transformation Details

Base Instruction: Please build upon the previous context or response rather than starting from scratch, maintaining continuity across the interaction.

Placement: prepend

Composition Behavior: accumulate

Parameter Effects:

  • reference:
  • When set to last: Build specifically upon the most recent response or context provided.
  • When set to specific: Build upon the specific part of the previous context mentioned in the current prompt.

  • When set to all: Consider all previous exchanges in the conversation when building your response.

  • approach:

  • When set to extend: Add new information, examples, or dimensions that extend and complement the existing content.
  • When set to refine: Improve, correct, or enhance the existing content while preserving its core message.
  • When set to contrast: Provide alternative perspectives or approaches that contrast with those in the existing content.

  • When set to synthesize: Combine and integrate ideas from the existing content into a new cohesive whole.

  • preserveStructure:

  • When set to true: Maintain the same organizational structure, format, and overall framework as the referenced content.
  • When set to false: Feel free to reorganize or restructure the content in a new way that better serves the current purpose.

Compatibility

  • Requires: None
  • Conflicts: None
  • Compatible Models: gpt-4o, gpt-4-turbo
  • Standard Version: 1.0.0 - 2.0.0
  • Refine: Enhances BuildOn BuildOn and Refine can work together to iteratively improve content over multiple interactions
  • Contrast: Enhances BuildOn When using the contrast approach, BuildOn works well with decorators that provide multiple perspectives