Platform overview

Six layers.
No blind spots.

Most tools cover one layer. H7 closes the entire loop — from kernel observation to court-admissible attestation — including cryptographic proof of what it couldn't see.

01 Observe02 Understand03 Attest04 Respond05 Deceive06 Comply
01 · QII · eBPF kernel probe
Observe

What the agent does — not what it says

Prompt-injection filters and EDR tools operate on observable text or process names. H7 operates at the syscall layer via eBPF, where there is no abstraction to manipulate. Every execve, mmap, ptrace, and connect is captured with nanosecond timestamps and process-causal attribution — regardless of what the agent reported to the orchestrator.

The probe is a Rust no_std eBPF program. It cannot be observed by the monitored process, cannot be injected from userspace, and adds less than 0.4% CPU overhead under sustained production load.

eBPF / QIIRust no_stdsched_switch · execve · ptrace< 0.4% CPU
Learn more
h7-sensor · live stream
17:42:03.001 execve agent-runtime → python3
17:42:03.019 connect 10.0.0.1:443 → api.openai.com
17:42:03.044 execve python3 → /bin/bash
17:42:03.045 execve bash → curl
17:42:03.061 connect curl → 198.51.100.42:80
17:42:03.062 ─── anomaly: unexpected child chain ───
17:42:03.063 ptrace curl → agent-runtime
QII events: 847 │ drops: 0 │ cpu: 0.3%
02 · SII · Dual-channel ingestion
Understand

Intent meets behavior — one unified model

Kernel events (QII) tell you what happened. The semantic channel (SII) tells you what was supposed to happen. H7 correlates both: a legitimate tool call that spawns an unexpected child process is detected even if each half looks normal in isolation.

The SII gateway integrates with LangGraph, vLLM, and any MCP-compatible runtime. No model fine-tuning, no prompt modification, no content inspection. The agent's declared intent is signed and compared against its observed kernel behavior at every decision boundary.

Dual-channel SIILangGraph / vLLMMCP gatewayIntent vs. behavior
Learn more
h7-brain · correlation
SII tool_call search_web(query='IBAN routing')
QII execve curl https://api.search.dev/…
✓ intent matches kernel behavior
SII tool_call read_file('/etc/passwd')
QII open /etc/shadow (agent PID 4821)
✗ declared: /etc/passwd observed: /etc/shadow
→ INTENT_MISMATCH conf=0.97
SII latency: 1.2ms │ correlated: 100%
03 · hash_v2 ledger · Ed25519 .cal
Attest

Every event sealed — including the ones we missed

Every kernel event is committed into a SHA-256 hash chain (hash_v2). The chain is sealed periodically with an Ed25519 signature. Any gap — whether from a sensor restart, a ring-buffer drop, or a deliberate blind-spot attack — is itself attested as a SENSOR_GAP event and sealed into the chain. You don't lose the proof of what you couldn't observe.

A BREACH episode produces a .cal certificate: a JSON+Ed25519 artifact that bundles the full incident timeline, the chain tip hash, and the behavioral evidence. Verifiable offline with a single public key. No CA, no SaaS, no network.

hash_v2 ledgerEd25519 .calBlindness attestationRGPD-safe pruning
Learn more
breach-20260619T174203Z.cal
{
"episode_id": "ep_9f3a2b…c01e",
"kind": "INTENT_MISMATCH",
"conf": 0.97,
"chain_tip": "a3f7c9…1d82",
"epoch": 14,
"sealed_at": "2026-06-19T17:42:03Z",
"signature": "Ed25519:9e4d…b7f1"
}
$ h7ctl verify breach-20260619T174203Z.cal
✓ signature valid ✓ chain continuous
04 · Muraille LSM · Kernel veto
Respond

Stop the action before the syscall completes

Detection without enforcement is telemetry. The Muraille LSM layer (Linux Security Module) inserts a veto hook before a flagged syscall is allowed to proceed. The malicious action never completes — the agent is contained at the kernel boundary, not after the fact.

Enforcement operates on a signed contract profile: a calibrated behavioral envelope derived from observed nominal sequences. Any attempt to execute outside the envelope in production triggers a cryptographically-attested ATTEMPT_BLOCKED event and, optionally, a circuit-breaker freeze token.

LSM / MurailleKernel veto (pre-exec)Circuit breakerATTEMPT_BLOCKED
Learn more
muraille-lsm · enforcement
17:42:03.063 LSM hook: bprm_check_security
task=curl parent=bash(agent)
profile: not in allowed set
→ VETO syscall blocked (EPERM)
→ agent PID 4821 circuit-breaker: FROZEN
17:42:03.064 ATTEMPT_BLOCKED sealed in chain
episode ep_9f3a2b…c01e conf=0.97
blocked/hr: 1 │ false-vetoes: 0
05 · h7-deception/v1 · Canary beacons
Deceive

Trap attackers — including AI agents — at the credential layer

A fake .env file placed alongside real configuration contains synthetic credentials wired to a honey-sink. The moment an attacker (human or AI agent delegated to a compromised workspace) uses a canary credential outbound, the honey-sink records a hit — and the brain seals DECEPTION_TRIGGERED with confidence=1.0.

There are zero false positives by construction: a credential that was never issued to any real service cannot produce a legitimate hit. The first use is proof of compromise. The event is sealed into the ledger and certified with the same Ed25519 .cal format as any other breach.

Canary .env beaconDECEPTION_TRIGGEREDconf=1.00% FP by construction
Learn more
deception-sink · hit received
# canary .env (lure — never real)
PAYMENT_API_KEY=sk_live_h7cnry_8f3a…
WEBHOOK_URL=https://pulsaride.com/api/callback/…
── sink hit ───────────────────────────────────
beacon_id: cnry_payments_prod
source_ip: 203.0.113.77
token_fp: sha256:7c4e…a219
DECEPTION_TRIGGERED conf=1.0 FP=0%
.cal sealed → BREACH episode opened
06 · DORA · NIS2 · EU AI Act
Comply

One artifact. Three regulatory frameworks. Zero SaaS dependency.

The .cal certificate produced at every BREACH event is a self-contained compliance artifact: timestamped kernel timeline, behavioral evidence, Ed25519 signature, and chain integrity proof. It satisfies DORA Art. 17 incident reporting, NIS2 Art. 21 supply-chain monitoring, and EU AI Act Art. 9 runtime audit-trail requirements in a single file.

Verification requires no network access, no external CA, no subscription. The Ed25519 public key is the only dependency. Auditors can replay every .cal bundle years after an incident without calling home.

DORA Art. 17NIS2 Art. 21EU AI Act Art. 9Air-gap verify
Learn more
compliance-export · audit bundle
$ h7ctl export --period 2026-Q2 --format dora
episodes: 3 (2 BREACH, 1 ATTEMPT_BLOCKED)
.cal files: 3 (Ed25519 verified)
chain gaps: 0 (continuous)
DORA Art.17 ✓ incident report ready
NIS2 Art.21 ✓ supply-chain evidence
AI Act Art.9 ✓ runtime audit trail
output: ./h7-audit-2026-Q2.zip (1.2 MB)

Ready to close the loop?

Deploy all six layers in a 6-week pilot — production or staging, H7 in your environment.

Apply for a Pilot →Try the Demo Kit