Sieve – scans Cursor/Claude chat history for leaked API keys
AI coding tools are leaking your API keys into plaintext databases. Sieve just plugged the gap.
Why it matters
As AI agents (Cursor, Claude Code, Copilot) become standard dev tools, they're creating a blind spot in security: chat histories and state files store unencrypted secrets outside traditional git scanning. Sieve addresses a real vulnerability in the AI-native dev workflow.
The key facts
5 to knowAI coding tools (Cursor, Claude Code, Copilot, Cline) embed API keys in unencrypted local SQLite databases during normal operation
Standard secret scanners (gitleaks, detect-secrets) only scan git repos — not AI transcript stores
Sieve product: macOS app that scans local chat history files, flags exposed keys by severity, redacts in-place, stores fingerprints in Keychain
Covers: Cursor, Claude Code, Claude Desktop, Copilot, Cline, Roo Cline, Windsurf, Gemini CLI, .env files
Published May 2026 on Hacker News (nascent product/tool announcement)
Go to the source
Hacker Newsapps.apple.com
Publisher excerpt: Background: I was using Cursor to set up an OpenAI integration.The agent read my .env file, added the key to the config, and everything worked. What I didn't think about: that key was now sitting in a plaintext SQLite database at ~/Library/ApplicationSupport/Cursor/User/workspaceStorage/.. AI…