GitHub ↗
Flagship · Rust · eBPF

Talus Process Monitor — behavioral ransomware detection that answers with a kill signal.

talus-process-monitor hooks syscall tracepoints in the Linux kernel, scores per-process open-rates in one-second windows and terminates the encryptor with SIGKILL — guardrailed, whitelist-aware, and never at the cost of the host. One static binary. No kernel modules. No telemetry.

Try 14 days free Deployment guide Source ↗

$50 per machine per year · fleet tier from $30 · one key per machine, delivered by email on payment · VAT invoice · or request a quote by email

~280k
events per second, single core
7.6%
sustained CPU under synthetic load
<1s
verdict to SIGKILL, same second
188
tests in the CI suite, every push
What it does

Three layers, one binary.

Detection is the easy half. Talus ships the half nobody writes up: the response path.

Detect: behavior, not signatures

Mass file-opens scored per process in 1s windows; canary reads, mass-delete and exfiltration-spike detectors. Signature-free — catches strains nobody has named yet.

Respond: SIGKILL in the same second

When the verdict fires, the offending process is killed immediately. Guardrails, per-path whitelists and canary verification keep it fast without being reckless.

Contain: Landlock & seccomp self-sandboxing

The agent sandboxes itself before it starts watching: no kernel modules, no writable footprint outside its own directory, verified by its own tests.

Ship: SIEM, Kafka, ClickHouse, MemGraph

Forward alerts as RFC5424 or CEF, stream raw events to Kafka, persist to ClickHouse, explore process lineage in MemGraph. Self-hosted, your infrastructure.

Sixty-second start

From pip to live dashboard.

pip install talus-process-monitor
talus-monitor install
sudo talus --web 10.0.0.5:8443
talus-monitor selftest --canary

The canary self-test simulates a ransomware burst with harmless file operations and prints the real verdict — detection proof on your hardware in under a minute. Full flags and tuning: /docs and the Field Guide.

Built for production

Real-world guardrails, real-world results.

Talus runs where attacks happen: inside the kernel, on your hosts, on your terms.

eBPF-native

Kernel-level process events via eBPF. No kernel modules, no reboot, near-zero overhead.

Privacy-first

Runs entirely on your servers. No cloud dependency, no telemetry, no third-party servers.

Instant activation

Buy through Stripe → activate in under 5 minutes. Ed25519-signed keys, server-side seats.

Active development

Ship every week. See the changelog and documentation.

Three minutes, three steps

From install to kill-switch.

1

Install

pip install talus-process-monitor && talus license activate <KEY> — eBPF probes attach to the kernel in seconds, zero config.

2

Activate

One key per machine. Seats enforced server-side. 30-day Enterprise trial on first run — no card required.

3

Monitor

Real-time process events, anomaly alerts, SIGKILL under 1s. Fleet dashboard at your domain.

Why Talus?

Standalone Linux, not Kubernetes.

Talus monitors individual servers and small fleets — no sidecars, no cluster, no Helm.

FeatureTalusbpftraceFalcoTetragon
eBPF kernel events✅✅✅✅
Process tree visualization✅❌⚠️⚠️
Fleet dashboard✅❌⚠️⚠️
Anomaly detection✅❌✅✅
No Kubernetes required✅✅❌❌
Standalone (no sidecar)✅✅❌❌
Setup time5 minCLI only30+ min30+ min
Price / year$50FreeFree–$500+Free–$500+
Privacy (no cloud)✅✅⚠️⚠️

Comparison based on publicly available documentation as of 2026. Talus targets standalone Linux servers, not Kubernetes-native security.

Questions

Frequently asked.

How is Talus different from Falco or Tetragon?

Talus is built for standalone Linux servers, not Kubernetes. No sidecars, no cluster, no Helm charts — deploy one binary and monitor.

What are the system requirements?

Linux kernel 5.8+ with BTF enabled. Ubuntu 20.04+, Debian 11+, RHEL 8+, Arch. Root for eBPF attachment. Minimal RAM/CPU overhead.

How does licensing work?

Single: one server, $50 / 200 PLN per year. Fleet: per-node, minimum 10 nodes ($30 / 117 PLN per node/year). Annual subscription, instant activation, cancel anytime.

Can I try before buying?

Yes — a 30-day Enterprise trial starts automatically on first run, no credit card. Need longer? Ask for 14 more days.

Is my data sent to the cloud?

No. Talus runs entirely on your servers. The optional fleet dashboard aggregates from your own nodes — no third-party servers.

What payment methods do you accept?

Stripe checkout — card payments globally, priced in USD or PLN.

Pricing

Priced like a tool, not like a platform.

$50 / machine / year

Enterprise tier — one key per machine, one year of updates.

  • Web dashboard, REST API, WebSocket event stream
  • Auto-kill response engine with guardrails
  • SIEM forwarding (RFC5424 / CEF), Kafka, ClickHouse, MemGraph
  • One key per machine — a key activates on exactly one host
  • License key delivered by email the moment payment settles
  • Email support — 48h business-day response
  • Detection engine stays MIT, forever
$30 / machine / year

Enterprise Fleet — the same features, minimum 10 machines.

  • Everything in Enterprise, per machine
  • Minimum 10 machines — set the quantity on the checkout page
  • Each machine gets its own key, issued automatically on payment
  • One subscription, one invoice, cancel or change size any time
10 machines or more: $30 / machine / year
$150 / session, billed in USD

Support Session — two hours with the engineer who wrote it.

  • Screen-shared, recorded on request
  • Written follow-up: config diffs, tuning values, exact commands
  • 14 days of email follow-up on the same topic
  • Tuning detectors & response guardrails for your workload
  • Air-gapped install, or lab → production migration

Component licensing

Detection engine (eBPF, scoring, sweep detectors)MIT — free
Dashboard, REST API, WebSocketper machine
Auto-kill response pathper machine
SIEM / Kafka / ClickHouse / MemGraph sinksper machine
SourcePublic, MIT

MSPs & resellers

Fleet tier starts at $30 per machine for 10 or more — your margin above that. Write to important@hartwell-labs.pl for the vendor order form — net-30 available on direct invoice.

Supply chain

Verifiable end-to-end.

Every release ships with an SBOM and build attestation; provenance is continuous, not advertised in slide decks.

SBOM (SPDX)

Full dependency inventory attached to every release asset.

Build attestation

Sigstore-style provenance linking source commit to published binary.

OpenSSF Scorecard

Supply-chain posture measured continuously on the flagship repository.

Verify a release Releases ↗
Email

Get eBPF & Linux security insights

Monthly. No spam. Unsubscribe anytime.

🛰️ Monitor your Linux servers in real time