Connect Your Agent to Any Tool

Your AI agent can call 40+ external APIs without ever seeing the raw credentials. Store your keys once, and the ClawTrust credential proxy handles authentication on every request.

How it works

Three steps between your credential and a secure API call.

1

Store your credential

Add your API key, OAuth token, or email login through the ClawTrust dashboard. Every credential is encrypted with AES-256-GCM before it touches the database.

2

Agent calls through the proxy

When your agent needs to call an external API, it sends the request to the ClawTrust credential proxy. The agent never sees the raw credential.

3

Proxy injects auth and forwards

The proxy decrypts the credential, attaches it to the outbound request (as a header, query param, or OAuth signature), and forwards it to the target API. The response flows back to your agent.

Supported credential types

Eight authentication methods covering virtually every SaaS API.

API Key (Bearer)

Standard Bearer token authentication. The proxy adds an Authorization: Bearer header to every request.

Examples: Stripe, OpenAI, HubSpot, SendGrid, VAPI, Housecall Pro

API Key (URL Parameter)

Appends the API key as a query parameter on each request URL.

Examples: Cal.com, legacy REST APIs

API Key (X-API-Key)

Sends the key in a custom X-API-Key header. Common in newer API designs.

Examples: Anthropic, custom APIs

OAuth2 Client Credentials

Machine-to-machine OAuth flow. The proxy handles token exchange and refresh automatically.

Examples: ServiceTitan, Salesforce M2M, QuickBooks

OAuth 1.0a (4-key signing)

Full OAuth 1.0a request signing with consumer key, consumer secret, access token, and token secret.

Examples: X (Twitter)

Email Login (Send/SMTP)

SMTP credentials for sending email as your agent. Supports app passwords and OAuth tokens.

Examples: Gmail, Outlook, Yahoo

Email Inbox (Read/IMAP)

IMAP credentials for reading incoming email. Your agent can monitor inboxes and respond to messages.

Examples: Gmail, Outlook

Username & Password

Basic auth or form-based credentials for services that use username/password pairs.

Examples: Jira, Twilio

Integration guides

Step-by-step setup instructions for each service.

How credentials are protected

Five layers of protection between your keys and the outside world.

AES-256-GCM encryption at rest

Every credential is encrypted before it reaches the database. Decryption only happens inside the proxy at request time.

Never stored on the agent's server

Credentials live on the ClawTrust control plane, not on your agent's dedicated VPS. The agent only has a proxy token.

Allowed domains restrict API access

Each credential has a list of allowed domains. The proxy rejects any request to a domain not on the list, preventing misuse.

Every proxy call is audit-logged

Full request logging with timestamp, target domain, HTTP method, and status code. Review your agent's API activity anytime.

Content scanning blocks exfiltration

Outbound responses are scanned for credential patterns. If your agent tries to leak a key in a chat message, it gets blocked.

Ready to connect your agent to your tools?

Store your credentials once. Your agent handles the rest, securely.