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
I have been using context-mode now for at least a week at this point, and I noticed yesterday that Claude had started creating files outside of the workspace/git-repository by having context-mode running commands that created said files.
The guard rails that exists in Claude Code are built around cc being able to identify certain patterns in the command the LLM wants to run, but these are "hidden" from cc when they are run through Context-Mode, and constantly having to allow every command that context-mode is good for makes for a very slow and "click-heavy" experience.
I'm currently trying to implement some context-mode specific guard rails, and if I am successful, I'll send a PR for it. But I wanted to bring some light to this issue as it is a pretty severe security issue.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
-
I have been using context-mode now for at least a week at this point, and I noticed yesterday that Claude had started creating files outside of the workspace/git-repository by having context-mode running commands that created said files.
The guard rails that exists in Claude Code are built around cc being able to identify certain patterns in the command the LLM wants to run, but these are "hidden" from cc when they are run through Context-Mode, and constantly having to allow every command that context-mode is good for makes for a very slow and "click-heavy" experience.
I'm currently trying to implement some context-mode specific guard rails, and if I am successful, I'll send a PR for it. But I wanted to bring some light to this issue as it is a pretty severe security issue.
Beta Was this translation helpful? Give feedback.
All reactions