PPhying中文
Curated security research, vulnerabilities, advisories and tools for practitioners.

Insecure credential storage plagues MCP

blog.trailofbits.com | blog | #mcp | #secrets-management | #llm-security | #vulnerabilities | #trail-of-bits | #credential-storage | #api-keys
📄 Raw .md⬇️ Download .md🔗 Original ↗

Summary

This post describes how many examples of MCP software store long-term API keys for third-party services in plaintext on the local filesystem, often with insecure, world-readable permissions.
Published
Collected

Related coverage

blog · blog.trailofbits.com
State divergence enables unauthorized access
This coverage gives security teams current context for monitoring, validation, and remediation.
blog · hackerone.com
iOS App Secret Management: Best Practices for Keeping Your Data Secure
Best practices for managing iOS app secrets: avoid hardcoding API keys, use .xcconfig build configurations, store runtime credentials in the Keychain, and prefer OAuth with short-lived tokens.
blog · wiz.io
Build AI Security Agents with Wiz MCP
Wiz MCP is now GA, letting AI assistants and custom agents query the Wiz Security Graph, reuse Red, Blue and Green agent analyses, and run AI Skills workflows for triage and remediation.
blog · blog.trailofbits.com
Your AI agent fixed the bug, but what else did it change?
We’re publishing patching-bugs and regression-hunt, two plugins that let engineers delegate patch development and regression checks to agents.
blog · blog.trailofbits.com
Auditing in the age of (good enough) AI
Trail of Bits explains how agents built tooling and a Lean model for the Miden zkVM audit, finding a flaw enabling Falcon signature forgery and yielding 95 machine-checked proofs.
blog · blog.trailofbits.com
1Password's AI patching benchmark is misleading
Trail of Bits says 1Password's FLAWED benchmark misstates AI patch quality, mixing wrong-fix prompts and no-test trials into a 26% clean-fix headline, and ships two agent skills for better fixes.
Back