You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: Documentation/ai_scanning.md
+2Lines changed: 2 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,9 +1,11 @@
1
1
# AI scanning requirement
2
2
3
3
In addition to a regular human review, review providers are expected to use AI (an LLM) to perform a security review with the issues triaged by the review provider. The review must be performed using the latest version (at the time the review starts) of one of the following models:
4
+
4
5
* Anthropic Opus
5
6
* Gemini Pro
6
7
* OpenAI GPT codex
8
+
7
9
This list will be regularly updated.
8
10
9
11
The review must be performed by a suitable harness for orchestration. Without an appropriate harness a review often goes off-track and does not look at all files. For now, the review provider may choose something they consider suitable, including their own tools. We can recommend [Arm Metis](https://github.com/arm/metis). We expect that over time more tools will be publicized and we will update this recommendation.
0 commit comments