SECRETBRIDGE · SECURITY

Clear boundaries. Verifiable trust.

Understand credential use, authorization checks, bounded results and explicit security limits.

LOCAL · EXPLICIT · INSPECTABLE

Credential use stays inside the broker.

Storage, authorization decisions and returned results each have a defined responsibility.

Native credential store

Passwords, tokens and private keys are not MCP arguments. The broker resolves credential references internally; public interfaces do not expose secret reads or exports.

Local authorization

Management sessions, IPC and MCP have separate boundaries. Approval decisions are not exposed to MCP; optional authentication accepts a user-provided code for an identified request.

Version and result constraints

Approval snapshots bind targets, configuration and lifetime. Output is filtered before return and storage; cancellation, revocation and resource changes trigger checks.

Limits you need to evaluate

  • A process with arbitrary code execution under the same OS account may bypass application approval or access the native store directly.
  • Filtering is not an OS sandbox and cannot identify every piece of business data or transformed secret.
  • Review controlled-terminal authorization carefully. Keep SSH host-key and TLS verification enabled.
  • Before installation, check the GitHub release source, file digest, package manifest and corresponding source.

Review validation by version

Platform support and validation scope can vary by release. Before installation, review the support matrix and release notes for your selected version, distinguishing automated checks, UI tests and native desktop acceptance.

Support matrix and evidence · Full security model

Report vulnerabilities privately

Use GitHub private vulnerability reporting. Public issues must not contain secrets, PINs, recovery keys, pairing material or sensitive logs.

Submit a private security report

Product content based on SecretBridge 0.3.1 · a82cd12 · Reviewed 2026-10-12