fix(security): prevent prompt injection in llm_validator#2066
Open
Waqar53 wants to merge 1 commit into567-labs:mainfrom
Open
fix(security): prevent prompt injection in llm_validator#2066Waqar53 wants to merge 1 commit into567-labs:mainfrom
Waqar53 wants to merge 1 commit into567-labs:mainfrom
Conversation
Fixes 567-labs#2056. - Add _format_validation_prompt() with XML-style escaping to prevent prompt injection attacks. User input is now wrapped in <user_value> tags to clearly separate it from instructions. - Add escape_user_input parameter (default: True) for backward compatibility when needed. - Fix allow_override logic bug: now correctly returns fixed_value when validation fails instead of only when valid. - Add async_llm_validator() for async validation pipelines. - Update exports in validation/__init__.py and instructor/__init__.py. - Add comprehensive unit tests for security improvements. Security Note: This prevents attackers from crafting inputs that could manipulate the LLM validator into always returning is_valid=true.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #2056.
Add _format_validation_prompt() with XML-style escaping to prevent prompt injection attacks. User input is now wrapped in <user_value> tags to clearly separate it from instructions.
Add escape_user_input parameter (default: True) for backward compatibility when needed.
Fix allow_override logic bug: now correctly returns fixed_value when validation fails instead of only when valid.
Add async_llm_validator() for async validation pipelines.
Update exports in validation/init.py and instructor/init.py.
Add comprehensive unit tests for security improvements.
Security Note: This prevents attackers from crafting inputs that could manipulate the LLM validator into always returning is_valid=true.