Noma Labs tricked GitHub's AI agent into leaking private repositories with a crafted issue. The prompt-injection flaw, GitLost, has no code fix.

A prompt injection attack can trick GitHub’s preview Agentic Workflows into retrieving content from private repositories and ...

Research shows signed Git commits can be re-encoded into fresh GitHub Verified hashes without changing files or breaking ...

A critical prompt injection vulnerability in GitHub Agentic Workflows could allow unauthenticated attackers to leak private repository data, ...

The IDE ruled when developers did one thing at a time. Now we need a next-generation tool to manage multiple coding agents ...

An indicator is a claim, not a fact — and the advisory covering your own network is the one you can least afford to accept ...

In the brief history of AI security, the prompt injection has quickly become the top threat. Large language models are ...

Hackers can trick GitHub's AI agent into leaking private code by posting fake issues in public repositories owned by the same organization. The flaw, called GitLost, exploits how AI agents treat user ...

Most widely cited AI coding benchmarks, including the original SWE-bench, were built primarily around Python repositories, meaning headline performance results may not accurately predict how coding ag ...

Discover the AI Agent Safety Directory that evaluates AI agents before deployment with trusted safety ratings, risk insights, and performance benchmarks.

The flaw allows an unauthenticated attacker to craft a GitHub Issue in an org's public repository and then silently pull data ...

The issue affects GitHub Agentic Workflows setups that read public input, hold private repo access, and can post output ...