10 hours ago
AI coding tools are useful, but troubleshooting can quickly turn into sharing more context than intended: source code, logs, internal URLs, API keys, or configuration files.
Where do you draw the line between useful context and information that should never leave your environment?
Do you sanitize every prompt, use local models, restrict what the assistant can access, or avoid these tools for sensitive projects? And when an AI agent can read files and run commands, how much access is too much?
Interested in practical approaches and common mistakes—keep examples anonymized and leave out actual secrets.
Poll: How do you handle sensitive projects when using AI coding tools?
- Cloud tools, with sanitized inputs and restricted access
- Local models only
- I avoid AI tools for sensitive projects
- I haven't established a clear policy
Where do you draw the line between useful context and information that should never leave your environment?
Do you sanitize every prompt, use local models, restrict what the assistant can access, or avoid these tools for sensitive projects? And when an AI agent can read files and run commands, how much access is too much?
Interested in practical approaches and common mistakes—keep examples anonymized and leave out actual secrets.
Poll: How do you handle sensitive projects when using AI coding tools?
- Cloud tools, with sanitized inputs and restricted access
- Local models only
- I avoid AI tools for sensitive projects
- I haven't established a clear policy