Linux assessment and change tools for the agentic era

Assess and manage Linux changes, human or agent.

Hanalyx builds assessment tools, a control plane, and a spec compiler for assessing and changing Linux with recorded evidence.

“No Linux change should ever be unsafe, unauditable, or unreversible, whether a human or an AI made it. That sentence is our entire company.”
The Hanalyx teamOur founding principle

●The portfolio

Three tools. One trust chain.

OpenWatch is the platform. Kensa is the engine. Specter compiles the specs the other two are built from.

№ 01 · platform
OW

OpenWatch

The compliance OS.

A fleet eye, heartbeat, and control plane. Continuous posture, daily snapshots and trends, score-drop alerts, governance workflows, and evidence an assessor can follow, for any team running Linux at scale.

№ 02 · engine
Kn

Kensa

The change engine.

Assesses Linux configurations and applies changes through defined mechanisms. A single binary for scripts and pipelines, and the engine OpenWatch runs on.

№ 03 · toolchain
Sp

Specter

The spec compiler.

A compiler for specifications: the toolchain we use to build OpenWatch and Kensa, open to anyone who wants machine-enforced spec rigor for their own systems.

Experience across

DHSFBIHPEAT&T

For AI platform teams

Give your agents a production-safe API surface.

Letting an agent run ssh+sudo on your fleet is unbounded. Kensa offers defined mechanisms with recorded results. Review each plan and its rollback support before an agent applies it.

01
Bounded blast radiusReview each mechanism before applying it. Some changes support captured-state rollback; non-transactional operations are skipped by rollback.
02
Auditable by defaultScan findings record check evidence. Remediation envelopes and report snapshots are separate artifacts; verify their signing and export support for the release you use.
03
Rollback as a primitiveKensa attempts rollback for supported steps when validation fails. Partial apply and failed restoration require operator review; rollback is not universal.
04
Deterministic across hostsRuntime capability probes detect each host, so one rule works across RHEL 8/9/10, Rocky, and Alma without per-host scripting.

Evidence

Inspect the recorded evidence behind a change.

Scan evidence, remediation envelopes, and report exports differ in what they prove. Remediation envelopes verify across runs only when configured with a persistent signing key. Check the artifact and release before relying on a signature. The example shown here illustrates a record; it is not a verified export.

  • Mechanism + pre-state hash. Records the captured starting state
  • Change command + exit code. Records the command and reported exit code
  • Validation result + reason. Records the validation outcome
  • Rollback verification. Inspect the restoration result and any unresolved state
  • Multi-framework mapping. Check the rule mappings for the selected benchmark
illustrative-evidence.jsonILLUSTRATIVE SAMPLE
{
"schema_version": "v2",
"transaction_id": "00000000-0000-4000-8000-000000000001",
"rule_id": "ssh-disable-root-login",
"host_id": "sample.example.com",
"started_at": "2026-11-11T14:22:00.000000000Z",
"finished_at": "2026-11-11T14:22:03.000000000Z",
"pre_state_bundle": [
{ "StepIndex": 0, "Mechanism": "config_set", "Capturable": true,
"CapturedAt": "2026-11-11T14:22:00.400000000Z" }
],
"apply_steps": [
{ "StepIndex": 0, "Mechanism": "config_set", "Capturable": true,
"Success": true, "Detail": "applied", "Stranded": false, "Staged": false }
],
"validator_results": [
{ "Name": "sshd_t_syntax_check", "Passed": false,
"Detail": "sshd -t reported an error", "Evidence": "sample output" }
],
"rollback_results": [
{ "StepIndex": 0, "Mechanism": "config_set", "Success": true,
"Detail": "restored [post-state: matches pre-state]",
"PartialRestore": false, "Source": "inline",
"ExecutedAt": "2026-11-11T14:22:02.800000000Z" }
],
"decision": "rolled_back",
"severity": "high",
"framework_refs": [
{ "FrameworkID": "cis_rhel9_v2", "ControlID": "5.1.20" },
{ "FrameworkID": "nist_800_53_r5", "ControlID": "AC-6(2)" }
]
}

Why now

AI agents are about to make more production changes than humans ever did. That is fine, provided someone can answer one question: can you prove the change was safe, and undo it if it wasn't?

Assess changes and plan for recovery.

Put OpenWatch and Kensa in front of your fleet, and your agents.