- Newest
- Most votes
- Most comments
Based on the available documentation, the placeholder variables in Amazon Bedrock agent prompt templates are designed to work across different models, though specific model support may vary by variable. The documentation shows that certain variables like $ask_user_input_examples$ are supported by specific models such as Anthropic Claude Instant and Anthropic Claude v2.0.
However, the current documentation does not explicitly address support for Amazon Nova or newer Claude models (beyond Claude v2.0) regarding placeholder variable compatibility. The placeholder system itself - including pre-processing, orchestration, and post-processing variables - is a feature of the Bedrock agent framework rather than being model-specific, which suggests these variables should generally work across supported models.
For the most current information about placeholder variable support with Amazon Nova and newer Claude models, I recommend:
- Testing the placeholder variables directly with your target model to verify compatibility
- Checking for any updated documentation or release notes specific to these newer models
- Reaching out to AWS Support for clarification on model-specific placeholder support
The core placeholder functionality for action groups, knowledge bases, conversation history, and other standard variables should remain consistent, but model-specific variables (like few-shot examples) may have different support levels across model families.
Sources
Use placeholder variables in Amazon Bedrock agent prompt templates - Amazon Bedrock
Relevant content
- asked 4 months ago
- AWS OFFICIALUpdated 9 months ago
- AWS OFFICIALUpdated 4 days ago
