Prompt the user for a secret value via a native macOS dialog (hidden input), then store it. The value never passes through the model or the chat — it goes directly from the user
AI agents use input_secret to create or update resources in Claude Secrets — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Claude Secrets environment.
This tool writes a secret to the macOS Keychain store. While the description emphasizes the privacy-preserving input path (value never passes through the model), the net effect is creating/storing a credential. This is a Write operation.
From the tool's definition 'Prompt the user for a secret value via a native macOS dialog (hidden input), then store it' — the tool stores a new secret value into the keychain
Documented attack patterns abuse exactly the kind of access input_secret gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Claude Secrets, and nothing reaches the server without passing your rules. This is the rule we recommend for input_secret:
{
"version": "1",
"default": "deny",
"tools": {
"input_secret": {
"limits": [
{
"counter": "input_secret_rate",
"window": "minute",
"max": 30,
"scope": "grant"
}
]
}
}
}input_secret stays usable, but capped — an agent stuck in a loop can't make hundreds of changes a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Prompt the user for a secret value via a native macOS dialog (hidden input), then store it. The value never passes through the model or the chat — it goes directly from the user. It is categorised as a Write tool in the Claude Secrets MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Claude Secrets MCP server in PolicyLayer and add a rule for input_secret: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Claude Secrets. Nothing to install.
input_secret is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the input_secret rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for input_secret. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
input_secret is provided by the Claude Secrets MCP server (vaultry/claude-secrets). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Claude Secrets, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.
Free to start. No card required.
6 Claude Secrets tools catalogued and risk-classified — across an index of 43,000+ MCP servers.