Your agent just got a cryptographic identity. Teleport's LLM Proxy inspects every instruction before it executes.

Your agents are running with shared API keys and standing access.

One compromised agent. One prompt injection. Full blast radius.

Teleport just shipped the infrastructure layer every production agent stack is missing.

Beams public beta now includes LLM Proxy and Delegated Identity.

LLM Proxy sits between your agent and its inference endpoint.

Every request and response is inspected and written to an immutable audit log.

This is enforcement BEFORE instructions execute — not after damage is done.

Per-Beam allow lists control which agents reach which inference endpoints.

Delegated Identity gives each agent a scoped, cryptographic identity at runtime.

No shared credentials. No standing access. No inherited permissions.

Each agent gets exactly the infrastructure access its task requires — and nothing else.

Least-privilege access becomes a runtime property, not a configuration checkbox.

If an agent is compromised, the blast radius is bounded by design.

This runs on Firecracker microVMs with 2-second startup.

Zero secrets. Full audit coverage. Identity injected before the first line of code executes.

Your agents don't need more capabilities. They need boundaries.

Audit your agent identity model today.

If your agents share credentials or inherit broad service account permissions, you are one prompt injection away from a breach.

SOURCE: https://www.helpnetsecurity.com/2026/06/16/llm-proxy-and-delegated-identity/
VERIFIED: Help Net Security (June 16, 2026), GlobeNewswire Teleport press release (June 16, 2026), beams.run
SIGNAL: Agent runtime identity is becoming infrastructure, not an afterthought. Every production agent stack needs cryptographic identity and instruction-level enforcement.