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.
$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
Detection is the easy half. Talus ships the half nobody writes up: the response path.
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.
When the verdict fires, the offending process is killed immediately. Guardrails, per-path whitelists and canary verification keep it fast without being reckless.
The agent sandboxes itself before it starts watching: no kernel modules, no writable footprint outside its own directory, verified by its own tests.
Forward alerts as RFC5424 or CEF, stream raw events to Kafka, persist to ClickHouse, explore process lineage in MemGraph. Self-hosted, your infrastructure.
pip install talus-process-monitortalus-monitor installsudo talus --web 10.0.0.5:8443talus-monitor selftest --canaryThe 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.
Talus runs where attacks happen: inside the kernel, on your hosts, on your terms.
Kernel-level process events via eBPF. No kernel modules, no reboot, near-zero overhead.
Runs entirely on your servers. No cloud dependency, no telemetry, no third-party servers.
Buy through Stripe → activate in under 5 minutes. Ed25519-signed keys, server-side seats.
Ship every week. See the changelog and documentation.
pip install talus-process-monitor && talus license activate <KEY> — eBPF probes attach to the kernel in seconds, zero config.
One key per machine. Seats enforced server-side. 30-day Enterprise trial on first run — no card required.
Real-time process events, anomaly alerts, SIGKILL under 1s. Fleet dashboard at your domain.
Talus monitors individual servers and small fleets — no sidecars, no cluster, no Helm.
| Feature | Talus | bpftrace | Falco | Tetragon |
|---|---|---|---|---|
| eBPF kernel events | ✅ | ✅ | ✅ | ✅ |
| Process tree visualization | ✅ | ❌ | ⚠️ | ⚠️ |
| Fleet dashboard | ✅ | ❌ | ⚠️ | ⚠️ |
| Anomaly detection | ✅ | ❌ | ✅ | ✅ |
| No Kubernetes required | ✅ | ✅ | ❌ | ❌ |
| Standalone (no sidecar) | ✅ | ✅ | ❌ | ❌ |
| Setup time | 5 min | CLI only | 30+ min | 30+ min |
| Price / year | $50 | Free | Free–$500+ | Free–$500+ |
| Privacy (no cloud) | ✅ | ✅ | ⚠️ | ⚠️ |
Comparison based on publicly available documentation as of 2026. Talus targets standalone Linux servers, not Kubernetes-native security.
Talus is built for standalone Linux servers, not Kubernetes. No sidecars, no cluster, no Helm charts — deploy one binary and monitor.
Linux kernel 5.8+ with BTF enabled. Ubuntu 20.04+, Debian 11+, RHEL 8+, Arch. Root for eBPF attachment. Minimal RAM/CPU overhead.
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.
Yes — a 30-day Enterprise trial starts automatically on first run, no credit card. Need longer? Ask for 14 more days.
No. Talus runs entirely on your servers. The optional fleet dashboard aggregates from your own nodes — no third-party servers.
Stripe checkout — card payments globally, priced in USD or PLN.
Enterprise tier — one key per machine, one year of updates.
Enterprise Fleet — the same features, minimum 10 machines.
Support Session — two hours with the engineer who wrote it.
| Detection engine (eBPF, scoring, sweep detectors) | MIT — free |
| Dashboard, REST API, WebSocket | per machine |
| Auto-kill response path | per machine |
| SIEM / Kafka / ClickHouse / MemGraph sinks | per machine |
| Source | Public, MIT |
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.
Every release ships with an SBOM and build attestation; provenance is continuous, not advertised in slide decks.
Full dependency inventory attached to every release asset.
Sigstore-style provenance linking source commit to published binary.
Supply-chain posture measured continuously on the flagship repository.