Gazebo
    ServicesAgentsDocsSpecWritingPricing
    Log inSign up
    Log in

    Getting Started

    • Introduction
    • How Gazebo works
    • Core concepts
    • Quickstart

    Agent Access

    • Access profiles
    • Bearer tokens
    • Audit logs

    MCP

    • Overview
    • Cursor
    • Claude Code
    • Windsurf
    • Any MCP client
    • get_credential reference

    Open Standard

    • AIP Specification ↗

    Sharing Agents

    • Overview
    • Publishing an agent
    • The consent screen
    • Installing a shared agent
    • The identity stamp

    REST API

    • Authentication
    • Verification endpoint

    Security

    • Overview
    • Encryption model
    • Credential storage
    • Audit and compliance
    • Responsible disclosure
    Documentation

    Getting Started

    • Introduction
    • How Gazebo works
    • Core concepts
    • Quickstart

    Agent Access

    • Access profiles
    • Bearer tokens
    • Audit logs

    MCP

    • Overview
    • Cursor
    • Claude Code
    • Windsurf
    • Any MCP client
    • get_credential reference

    Open Standard

    • AIP Specification ↗

    Sharing Agents

    • Overview
    • Publishing an agent
    • The consent screen
    • Installing a shared agent
    • The identity stamp

    REST API

    • Authentication
    • Verification endpoint

    Security

    • Overview
    • Encryption model
    • Credential storage
    • Audit and compliance
    • Responsible disclosure
    DocsSecurityCredential storage

    Credentials are stored in the Gazebo vault — a PostgreSQL table where every value is encrypted at rest with AES-256-GCM before being written.

    Storage model

    Each credential entry stores:

    • user_id — which account owns this credential
    • service — the service name (e.g. stripe)
    • ciphertext — the AES-256-GCM encrypted credential value
    • nonce — the random nonce used for this encryption (required for decryption)
    • created_at, updated_at

    The plaintext credential value is never written to the database.

    Zero-plaintext guarantee

    After a credential is stored, its plaintext value is never returned to any web client. This applies to:

    • The Settings page in the dashboard (shows connection status, not the credential)
    • The API (no endpoint returns plaintext credentials to authenticated web sessions)
    • Error messages and logs (credentials never appear in plaintext in any log)

    The only context where decryption occurs is server-side, during an authorized get_credential call from an agent with a valid bearer token and the correct service in its access profile.

    What "connecting a service" does

    When you connect a service:

    1. You paste your API key or complete OAuth
    2. The server receives the plaintext value over TLS
    3. The server encrypts it with AES-256-GCM using the vault key
    4. The ciphertext and nonce are written to the database
    5. The plaintext is discarded from memory

    There is no "view credential" feature. If you need to see a stored credential value, you must retrieve it via get_credential using a valid agent token — which is logged.

    Encryption key management

    See Encryption model for key details. The critical operational constraint: the VAULT_ENCRYPTION_KEY must never change after setup. Key rotation requires re-encrypting every stored credential — a manual migration that must be planned carefully.

    OAuth credentials

    For OAuth-connected services (Stripe OAuth, Google OAuth, etc.), Gazebo stores both the access token and the refresh token in the vault. Refresh tokens are re-encrypted if rotated. The OAuth flow never exposes tokens to the browser after the initial connection.

    Explore Gazebo

    Gazebo gives each agent an independent access boundary around your agent credentials, with controls for approval, auditing, and revocation.

    Scoped identities for AI agentsGive every agent only the credentials it needs.Credential security and audit controlsSee how access is protected, recorded, and revoked.
    Encryption modelAudit and compliance
    Gazebo

    IAM for AI agents. Scoped credentials, access policies, and audit trails — without rotating keys.

    Product

    • Pricing
    • Status

    Explore

    • Services
    • Agents
    • Workflows
    • Integrations

    Content

    • Writing
    • Topics
    • Blog
    • Docs

    Free Tools

    • Scanner

    Company

    • About
    • hello@gazebohq.com
    • security@gazebohq.com

    © 2026 Gazebo. All rights reserved.

    PrivacyTermsSecurity