Conversation
… validation Training data cutoff can cause the agent to be suspicious of CVEs from years beyond its training data (e.g., CVE-2026-* when training ended in 2025). Add explicit instruction to check the system date before dismissing CVE identifiers as invalid. Co-authored-by: openhands <openhands@all-hands.dev>
all-hands-bot
approved these changes
Mar 21, 2026
Contributor
all-hands-bot
left a comment
There was a problem hiding this comment.
🟢 Good taste - Simple, pragmatic fix to a real problem. Agents shouldn't flag valid CVEs just because they're from years beyond training data. Checking the system clock is the right approach.
all-hands-bot
approved these changes
Mar 21, 2026
Contributor
all-hands-bot
left a comment
There was a problem hiding this comment.
🟢 Good taste - Simple instruction that solves a real problem: agents shouldn't flag valid CVEs just because the year exceeds their training cutoff. Checking the system clock is the right pragmatic fix.
aivong-openhands
pushed a commit
to OpenHands/software-agent-sdk
that referenced
this pull request
Mar 22, 2026
This change helps agents recognize that their training data may have a cutoff date earlier than the current system clock. When evaluating time-sensitive information like CVE identifiers (e.g., CVE-2026-*), the agent should trust the system clock rather than making assumptions based on training data. Closes OpenHands/extensions#116 Co-authored-by: openhands <openhands@all-hands.dev>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
The codereview-roasted skill can become suspicious of CVEs from years beyond the LLM's training data cutoff. For example, an agent trained on 2025 data might incorrectly flag CVE-2026-* identifiers as invalid or suspicious.
Solution
Added an explicit instruction in the Security and Correctness section telling the agent to check the system clock (
date) before evaluating CVE identifiers. This ensures the agent uses the actual current year rather than assuming based on training data.Changes