Latest writing
Does Your MCP Secret Server Hand the Plaintext to the Model?
Inspect MCP response schemas, credential injection, and output handling to see whether secret values can enter your AI agent's context.
Read the article ↗VaultIdeas worth
Ideas worth
keeping in mind.
Security architecture, everyday development, and the changing relationship between agents and secrets.
Is Phase Zero-Knowledge? E2EE and Server-Side Encryption
Phase uses end-to-end encryption by default and offers optional server-side encryption. Here is how those modes change the trust boundary.
Read the article ↗How to Keep API Keys Out of an AI Coding Agent
A practical guide to preventing API keys from leaking into Claude Code, Cursor, and Cline — covering common failure modes, vault patterns, and incident response.
Read the article ↗Apple Keychain Is Not Enough to Protect Secrets from Claude Code
macOS Keychain keeps secrets encrypted at rest. The moment you use the security CLI to pull them into your shell — and Claude Code is running — you've already lost.
Read the article ↗AI Agents On-Premise — Where Credentials Should Live
How parent environments, child-process injection, and credential brokers affect secret exposure in self-hosted AI workflows.
Read the article ↗What the Claude Code Source Leak Reveals About AI Agent Security
Anthropic accidentally shipped 60MB of Claude Code's unobfuscated source code in an npm package. Here's what security researchers found inside — and what it means for how you think about AI agent credentials.
Read the article ↗Cloudflare Workers Secrets and AI Development — What's Actually Safe
Separate deployed Worker secret bindings from local development credentials, and understand which processes and tools can read each value.
Read the article ↗Stop Telling Claude to Read Your .env File
The advice spreading across AI communities — 'just create a .env file and have Claude read it' — is wrong, and here's why it matters.
Read the article ↗How to Keep API Keys Secure When Using Claude Code
Keep API keys out of readable project files and use scoped command execution to reduce exposure in AI coding workflows.
Read the article ↗How to Store Secrets for Claude Code (Without Putting Them in .env)
The standard .env approach exposes your API keys to Claude's context window. Here's the correct way to store and use secrets with Claude Code.
Read the article ↗MCP Secret Manager — How OpaqueVault Works
How OpaqueVault encrypts secrets locally, injects them into commands, and keeps values and subprocess output out of MCP responses.
Read the article ↗MCP Server Secrets Management — The Complete Guide
How to manage credentials across MCP servers without exposing them to the AI. Architecture, patterns, and a reference implementation with OpaqueVault.
Read the article ↗Zero-Knowledge Secret Manager — What It Is and Why It Matters for AI Agents
A zero-knowledge secret manager means the server cannot decrypt your secrets. Here's why that architecture matters — especially when AI coding agents are in the loop.
Read the article ↗24,000 secrets found in MCP config files — what it means and how to fix it
Researchers found 24,008 live API keys and credentials inside Claude Desktop MCP configuration files. Here's what went wrong and how zero-knowledge vaults change the equation.
Read the article ↗HCP Vault Secrets End of Life — Migration Options
HCP Vault Secrets reached its announced end-of-life deadline in July 2026. Review replacement options and the requirements for rebuilding your workflow.
Read the article ↗Why your AI coding assistant is a secret leak waiting to happen
Claude Code, Cursor, and GitHub Copilot dramatically accelerate development — and, if misconfigured, dramatically accelerate credential exposure. Here's the structural problem and how to fix it.
Read the article ↗