// engagement · the layer under your code
Everything your product trusts before your code runs.
We read the image you ship, not the build flags you believe are set. No product subscription required.
23 components mapped, 15 analyzable
Containers named, not counted
Measured from the image, not the build
Standalone, no platform license
What we review
The seam between silicon, firmware and software, which is where a product's real trust decisions live.
- The boot chain, link by link. What verifies what, with which key, and what happens on failure. A chain whose every link verifies and whose recovery mode does not has exactly one link.
- The signature check as implemented. Not whether verification exists, but whether the parsing before it is attacker-reachable, whether rollback is accepted, and whether the failure path is fail-closed.
- Attestation and key handling. Where the root of trust is, what it measures, and whether a measurement anybody can influence is treated as evidence.
- The update and recovery path, which is the most common way a hardened product gets owned: downgrade, partial write, interrupted update, factory reset, debug mode.
- The components nobody scoped: the embedded controller, the BMC, option ROMs, Wi-Fi and Bluetooth firmware, the retimer.
- The mitigation inventory, read out of the binary rather than from a build flag somebody believes is set.
What you get back
Boot chain map
Link by link, with the gaps
Findings
Confirmed and likely, kept apart
Reproductions
For anything confirmed
Component account
Opened, and not opened
Mitigation inventory
Read from the artifact
The fixes
Ranked by what breaks a chain
Where the review is feeding a regulatory file, the same evidence lands in the shape the CRA page describes, with the obligation and its limit in the same row.