Trusted Execution Environment (TEE)

Integrity guarantees that enclave code and data cannot be tampered with or replayed by a malicious OS or hypervisor. The hardware maintains a cryptographic Merkle tree over the entire enclave memory region. Any attempt to modify a cache line, swap it with stale data, or inject a fault is detected by the memory encryption engine during a fetch, triggering an immediate machine check exception that halts the enclave.

• Anti-Replay: Version counters prevent rollback to old, valid memory states.
• Anti-Splicing: Each cache line's address is bound to its MAC.

Sealing is the mechanism that allows an enclave to persist secrets to untrusted storage (disk) while maintaining confidentiality and integrity. The enclave encrypts data using a key derived from the processor's unique Root Seal Key fused in hardware. The sealing policy can bind the data to the enclave's identity (MRENCLAVE) or to the signing authority (MRSIGNER), allowing controlled data migration across software versions.

• Seal to Enclave: Only the exact same code version can unseal.
• Seal to Signer: Any enclave signed by the same authority can unseal, enabling upgrades.

The enclave's lifecycle begins with a measured launch, where the hardware cryptographically hashes every page of code and data loaded into the enclave before any instruction executes. This measurement, stored in the MRENCLAVE register, represents a cryptographically unique identity of the enclave's initial state. The launch is atomic—if any page fails to load or its hash mismatches, the enclave creation fails.

• Trusted Compute Base (TCB): The measurement captures the exact TCB.
• Reproducibility: Deterministic builds produce identical MRENCLAVE values.

A critical companion mechanism to TEEs that enables an agent to generate cryptographically signed proof of its current software stack and identity. This allows a remote verifier to establish trust before any interaction begins.

• Verifies the exact hash of code running inside the enclave
• Prevents man-in-the-middle substitution attacks
• Essential for establishing trust between agents from different administrative domains

A cryptographic protocol that allows a group of agents to jointly compute a function over their private inputs while keeping those inputs completely confidential from one another. Unlike TEEs, MPC does not require hardware trust assumptions.

• Complements TEEs for cross-organizational agent collaboration
• Provides information-theoretic security in some configurations
• Higher computational overhead than TEE-based approaches

A cryptographic method enabling one agent to prove to another that a statement is true without revealing any information beyond the validity of the statement itself. ZKPs and TEEs are often combined for defense-in-depth.

• TEEs provide execution privacy; ZKPs provide verifiable computation
• Enables agents to prove correct execution without revealing inputs
• Used in private smart contracts and anonymous credentials

The isolated execution region within a processor that forms the core of a TEE. Enclaves maintain memory encryption and integrity protection even against a compromised operating system or hypervisor.

• Intel SGX creates enclaves with up to 512MB protected memory
• AMD SEV encrypts entire virtual machines
• ARM TrustZone splits the processor into secure and normal worlds

The property of a distributed system to reach consensus and continue operating correctly even when an arbitrary number of nodes fail or act maliciously. TEEs strengthen BFT by ensuring individual agents execute logic faithfully.

• Prevents equivocation attacks where agents send conflicting messages
• TEE-backed BFT can tolerate up to f faulty nodes out of 3f+1 total
• Critical for agent voting and consensus protocols