Today we're introducing the PTERI MCP Connector, bringing PTERI Wallet's identity and authorization model directly into the Model Context Protocol ecosystem. Any MCP-compatible client can now authenticate against a user's PTERI identity using the same signature-based model that secures human sign-in, without provisioning a separate API key or shared secret for every agent.
What the connector does
AI agents are increasingly acting on people's behalf — reading calendars, moving data between tools, calling other agents. Most of that access today is secured with long-lived API keys or shared secrets sitting in configuration files, where a single leak can expose everything an agent was ever allowed to touch.
The PTERI MCP Connector gives every agent its own cryptographic identity instead:
- Authenticate an agent the same way a person signs in — a challenge is issued, the agent's key signs it, PTERI verifies the signature. No password, no shared secret, nothing to phish.
- Delegate a scoped, time-boxed token from a user's wallet to a specific agent, rather than handing over standing access.
- Revoke a single agent's access instantly, without rotating credentials for every other agent connected to the same account.
- Every authentication and delegation event is logged against the specific identity that performed it.
How it fits into an existing setup
The connector doesn't require replacing anything already in place. It sits alongside existing account authentication and applies specifically to the agents and tools a user chooses to connect — MFA continues to protect human sign-in, and the connector extends the same trust model to whatever acts on that person's behalf afterward.
Where it fits today
The connector is available for developers building on MCP-compatible platforms. As with any new integration, we'd recommend evaluating it in the PTERI Playground first before wiring it into a production environment.
We'll follow up here as the connector moves from beta toward general availability.