Secure handoffs for scripts, tools, and agents

Why developers buy it

Automation turns Shhhs from a manual secret sender into an operational control: repeatable handoffs for scripts, CI, support runbooks, and agent workflows with scoped identities.

• Repeatable handoffs
• Scoped automation identities
• Less plaintext in tickets and logs

Fixed Secret URLs

Give a script, CI job, or agent a stable Team alias such as github-deploy-key. The alias resolves the current encrypted version with a scoped API key, while the underlying secret still expires by TTL and views.

• Stable alias for runbooks
• Rotate encrypted versions
• Resolve only with fixed:read

API

Create and manage supported secret-sharing workflows from your application. Authentication, scopes, rate limits, lifecycle controls, and error behavior are documented in the API reference.

• Authentication
• Scopes
• Rate limits
• Lifecycle controls
• Error behavior

CLI

Use terminal workflows for repeatable delivery and retrieval. Agent-safe modes should support local sources and destinations without printing plaintext to standard output.

• Environment variables
• Files
• Dotenv entries
• Clipboard or file output
• No plaintext stdout

MCP-compatible workflows

Allow approved agent tools to coordinate a handoff while limiting where plaintext may appear. A workflow is only described as agent-safe when the deployed client, server, logs, and destination behavior have been verified.

• Approved agent tools
• Controlled handoff
• Verified client, server, logs, and destination behavior

Mobile-first private rooms

Private Rooms are designed to work as standalone links on mobile before native apps. A recipient can open only the room surface, not the full console, and the exchange remains bounded by plan limits and expiry.

• Standalone room links
• Token-to-token exchange
• Native apps can build on the same room contract

Security rules for automation

Automation should keep service credentials scoped, short-lived, quiet in logs, and separated from human login.

• Scope every service credential
• Use short expiration by default
• Avoid plaintext command-line arguments
• Do not print secrets to logs or model transcripts
• Rotate and revoke service credentials
• Keep automation identities separate from human login

FAQ