“Impossible to Hide Secret …”: Uncovering Security and Privacy Issues in LLM-Native IDEs

Summary

LLM-native IDEs (LIDEs) that allow AI agents to read repositories, execute commands, and modify code introduce significant security and privacy risks, as developers report incidents like leaked API secrets and file deletions. The authors analyzed 15.4 million Reddit posts from 46 subreddits using a three-stage pipeline, manually verifying 446 relevant posts and 6,280 comments for qualitative coding. They developed a taxonomy of 32 issues across 10 categories, with unauthorized file operations being the most frequent security problem (43.1%) and lack of transparency the top privacy concern (45.9%). Seven of the ten issue categories stem from system-level design choices rather than the underlying LLM’s reasoning. Developers commonly use self-devised workarounds such as configuration management (33%) and code governance (31%), indicating a trust gap between expectations and actual LIDE guarantees.

Takeaways

1. 1

   Cursor (34.5%) and Claude (23.7%) dominate Reddit discussions of LIDEs; unauthorized file operations are the most common security issue (43.1%), with destructive deletions accounting for 28.3% of security incidents.

2. 2

   System-level design choices—how the IDE grants access, executes actions, and manages context—cause 7 out of 10 issue categories, not the underlying LLM’s reasoning.

3. 3

   Developers most often rely on configuration management (33%, e.g., restricting permissions, disabling telemetry) and code governance (31%, e.g., manual review, version control as safety net), revealing a trust gap between expectations and actual LIDE guarantees.

FAQ