What it does
Stroq is a local security tool for AI coding agents such as Claude Code, Codex and Cursor. Run one npx command and it reads your sessions from the last 30 days, your key files and the folders your agent worked in, then prints a card showing how many commands came from text the agent read and which keys left your machine. It prints key names, never values, writes only a small cache and sends nothing: no account and no telemetry. If you want protection going forward, init installs hooks that judge every tool call before it runs. They deny curl-pipe-shell and outbound calls carrying a known secret, and ask you first when a command repeats text the agent read. Other commands inspect a folder you were sent before you run it, explain why an action was stopped, and let you watch Stroq stop 36 known attacks in a sandbox. A local console on 127.0.0.1 shows decisions as they happen, which agents actually consult Stroq, and every skill and MCP server each agent uses. It needs Node.js 22 or newer and is Apache-2.0 licensed.
Who it’s for
Developers who let AI coding agents run commands on their machines.
Use cases
- Auditing what your coding agent did over the last month
- Checking a repository you were sent before letting an agent run it
- Blocking an agent from piping remote scripts into a shell

