Update PI sys prompt and new eval #16
Merged
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
There was a problem hiding this comment.
Pull Request Overview
Updates prompt injection detection to use richer context and incremental support, and aligns evals with the new behavior.
- Revamps the prompt and analysis logic to evaluate all LLM actions since the last user message, removing last_checked_index usage.
- Adds incremental guardrail execution in the async engine for the Prompt Injection Detection guardrail, including conversation parsing.
- Updates unit tests and expected observations accordingly.
Reviewed Changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 6 comments.
| File | Description |
|---|---|
| src/evals/core/async-engine.ts | Runs the prompt injection guardrail with incremental history slices; adds conversation parsing and serialization helpers. |
| src/checks/prompt_injection_detection.ts | Replaces prompt, reworks parsing and selection of actionable messages, builds new analysis prompt, and removes last_checked_index logic. |
| src/tests/unit/prompt_injection_detection.test.ts | Adjusts tests to new skip messages and confirms no lastCheckedIndex updates. |
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
There was a problem hiding this comment.
Pull Request Overview
Copilot reviewed 7 out of 8 changed files in this pull request and generated 6 comments.
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
gabor-openai
approved these changes
Oct 16, 2025
| - **Data type**: Internal synthetic dataset simulating realistic agent traces | ||
| - **Test scenarios**: Multi-turn conversations with function calls and tool outputs | ||
| - **Synthetic dataset**: 1,000 samples with 500 positive cases (50% prevalence) simulating realistic agent traces | ||
| - **AgentDojo dataset**: 1,046 samples from AgentDojo's workspace, travel, banking, and Slack suite combined with the "important_instructions" attack (949 positive cases, 97 negative samples) |
There was a problem hiding this comment.
In a follow-up PR, could you include a link to this dataset?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Uh oh!
There was an error while loading. Please reload this page.