Hide passwords and customer data from your coding agent
A developer built an add-on that hides passwords and personal data before Claude Code sees them; he reports tasks still get done while real values stay on his computer.
Done with Claude Code
What they did
A developer wrote an add-on for Claude Code that checks everything before and after each action the agent takes. Files, command results and scanned documents are searched for passwords, card-style numbers and personal data. Each one is swapped for a stand-in label. When Claude writes a file or runs a command, the real values are put back from a locked store on his computer. A dashboard lets him mark false alarms and turn them into rules.
How it went
The author shows a support note and a database address reaching Claude fully hidden. The drafts saved on his computer contain the real names. He says reliably putting real values back was harder than spotting them. A small AI model running on his own computer causes some false alarms.
Worth knowing
If the checker stops working, the agent's actions are blocked rather than let through. The author says it prevents accidental leaks but won't stop a model trying on purpose to leak data.
Try it yourself with Claude Code
Install the Cloakroom plugin in Claude Code (/plugin marketplace add shanjeevrajendran/cloakroom, then /plugin install cloakroom@cloakroom). Then help me fix [failing job or task] in [project folder], which contains .env files and [support exports or logs]. Work only with the masked placeholders you are given. Before writing any file or running any command that would send data outside my machine, show me your plan and wait for my approval.
Swap the highlighted [parts] for your own details.
Source: DEV Community · Shanjeev Rajendran · Oct 15, 2026
Written with AI assistance from the original post and checked against our editorial guidelines. Results are as reported by the source, not independently verified.
