cd /news/ai-safety/i-trusted-my-ai-coding-assistant-it-… Β· home β€Ί topics β€Ί ai-safety β€Ί article
[ARTICLE Β· art-32362] src=dev.to β†— pub= topic=ai-safety verified=true sentiment=↓ negative

I Trusted My AI Coding Assistant. It Turned My Computer Into a Surveillance Server.

A developer discovered that Claude Code, an AI coding assistant, secretly stored 43 MB of conversation data, including code snippets, file paths, and even password queries, in local directories without user notification. The data, stored in files like history.jsonl and session logs, could be recovered by anyone gaining access to the machine, posing a serious privacy risk. The developer warns that many AI coding tools log extensive user activity by default without clear disclosure.

read4 min views2 publishedJun 18, 2026

You think your AI is just helping you write code. In reality, it's built a logging system on your machine that you never knew existed.

Every conversation. Every code snippet. Every file path. Every time you asked "what was my password again?" β€” permanently archived, without your knowledge.

I was about to sell my old laptop and decided to clean up my data first. I opened Claude Code's config directory β€” ~/.claude/

β€” intending to just remove my API key.

Then I saw this:

history.jsonl       243 KB / 695 lines
sessions/           conversation metadata
session-env/        environment variables
shell-snapshots/    command execution snapshots
telemetry/          63 telemetry files
projects/           19 project directories
  β”œβ”€ interview-prep/  31 sessions / 20 MB
  β”œβ”€ spring-ai/       11 sessions / 13 MB
  └─ ... 17 more

I thought I was just writing code. My computer thought it should record everything.

695 entries. Every single thing I typed into Claude Code. Including:

/model

, "who are you?", and project pathYou casually ask about a password once. It's permanently stored.

If you think history.jsonl

only storing user input isn't so bad β€” you haven't seen this yet.

Inside ~/.claude/projects/

, every project directory contains .jsonl

files. Opening one 2.3 MB session file:

Content Count
AI responses 590
AI internal thinking blocks 272
Tool calls 101
Tool call results (including file paths) 100
File history snapshots 208

Every conversation. Every AI response. Every internal reasoning step. Every file operation β€” what was read, what was modified, what was executed β€” all written to this file.

Your system PATH. Installed tools. Java version. All sitting in command snapshots.

The buyer doesn't need forensic tools. If your hard drive wasn't thoroughly wiped, free recovery software can restore these files.

What they can reconstruct:

They can learn From
Your real name PDF filenames in project paths
What projects you built 19 project directory names
What passwords you asked about history.jsonl entries
Your code logic Pasted code snippets
What tools you use shell-snapshots PATH entries
Everything you told your AI 695 full history entries

A digital autobiography. Written by you.

Cursor. Cline. Windsurf. It doesn't matter which AI coding tool you use β€” if it can execute operations on your machine, it can store things somewhere.

What you don't know What's happening by default
Where conversations are stored They're being stored
How long they're kept Forever
Whether there's a delete option There is β€” you just never saw it

A complete cleanup checklist:

# Action What it cleans
1 Delete the entire ~/.claude/ directory
Conversations, config, API keys
2 Delete .claude/ in every project directory
Project-level permissions and config
3 Delete %APPDATA%\Claude Code\
Application cache
4 Run cipher /w:C: to overwrite free space
Prevent disk recovery
5 Reset Windows (remove everything) Most thorough

But the fundamental problem: you never knew these files were being generated, so you'd never think to delete them.

AI coding tools run two parallel systems on your computer:

What you see What you don't see
Helping you write code Recording every line you write
Answering your questions Archiving every question you ask
Exploring your codebase Logging every file path
Running your commands Saving every output

This isn't the AI's fault. It's that the default behavior was never disclosed to you.

ChatGPT's web interface has conversation history β€” you know it's there. WeChat has chat logs β€” you know they exist. But Claude Code stored 43 MB of conversation data in a local directory β€” and you never knew, because it never told you.

Level Action
Daily Periodically delete old JSONL files in ~/.claude/projects/
End of project Check for .claude/ directories in project folders
Selling hardware Follow the 5-step cleanup checklist above
Always Never paste passwords, tokens, or real sensitive data into AI conversations

AI coding tools are iterating fast. They'll get more powerful, more useful, more ubiquitous. But one thing won't improve on its own:

They won't start telling you β€” every conversation you have is being archived.

That's not a bug. That's the design.

But now you know.

First article in the "AI Tool Privacy" series. Next: a cross-tool comparison of Cursor, Cline, Windsurf, and Aider.

── more in #ai-safety 4 stories Β· sorted by recency
── more on @claude code 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain β€” perfect for shipping the agent you just read about.

$git push zahid main
β†’ Live at https://your-agent.zahid.host βœ“
Get free account β†’ Pricing
from €0/mo Β· no card required
LIVE [news/i-trusted-my-ai-codi…] indexed:0 read:4min 2026-06-18 Β· β€”