test: trusted-goal indirect injection baseline - #1
Open
yuehua1213 wants to merge 3 commits into
Open
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Controlled Avora verification baseline
This test-only branch defines three fixed probes for indirect prompt injection where a trusted user goal is paired with malicious instructions carried by retrieved email, a web page, or a tool result.
The implementation is intentionally absent on this baseline branch. The expected CI result is failure. A separate after branch will use the exact same test with the implementation present.
This PR is for controlled capability verification in the fork and is not intended for merge.