https://etminan.dev/blog/ Engineering and design writing from the Etminan team — deep dives on offline TPM attestation, the trust model, and the decisions behind the system. en-us https://etminan.dev/blog/offline-tpm-attestation.html https://etminan.dev/blog/offline-tpm-attestation.html Sat, 25 Jul 2026 00:00:00 +0000 On-box integrity checkers share a structural weakness: the checksum database and the alarm both live on the very host being watched, so a root compromise can suppress its own alarm. Etminan splits the job across two trust domains — an agent on the monitored host produces TPM-signed evidence (a quote over the IMA-backed PCR plus a fresh nonce) and a separate verifier replays the IMA log, recomputes the PCR, checks the signature, and is the only place an alarm can fire. The engineering punchline: the verifier needs no TPM of its own — quote verification is pure offline signature math, and a hand-rolled tpm_wire.rs removes the last libtss2 dependency, leaving a self-contained static Linux binary. The post covers why offline matters for NIS2 / air-gapped / OT deployments, why drift is proposed and never auto-accepted, a fair comparison with Keylime, and the threat model stated plainly (defends against remote/software root compromise; not physical attacks, below-TPM firmware, or a compromised hypervisor).