ARTEX Defense Lab

v0.1.0 • Experimental detection engineering • Synthetic validation

A Sigma detection pack and offline purple-team test harness for ARTEX-relevant behavior. Includes seven behavioral rules, five correlations and two upstream Xint/Theori fingerprints, with fixtures, telemetry requirements and a validation playbook.

Original ARTEX source

We have cross-checked the existing fingerprints against the original ARTEX repository preserved by Software Heritage, pinned to revision e6ec56999175509551a1e43a3ee3b053d0941bd9. See the source mapping and limitations. This is not a claim of complete attack-chain coverage or live validation.

Quick start

Python 3.10+ required. From this repository:

python3 -m venv .venv
. .venv/bin/activate
python -m pip install -e '.[test]'
make test
artex-replay fixtures/positive.jsonl --root .
artex-replay fixtures/benign.jsonl --root .
artex-replay fixtures/known_false_positive.jsonl --root .

Replay reads JSONL and emits JSON matches without executing event content or contacting a target. Atomic matches provide context; filter on kind == "correlation" for correlation alerts. The positive fixture produces five alerts, the benign fixture produces none, and the approved credential-rotation fixture produces one known false positive.

Contents

  • vendor/theori/: two unmodified Xint/Theori Sigma user-agent fingerprints.
  • rules/atomic/: seven original, low-confidence context rules.
  • rules/correlation/: ordered web/auth/data and cloud identity sequences; distinct secret-read threshold.
  • src/: bounded offline reference evaluator, not a general Sigma engine.
  • tests/, fixtures/: parser, reference, logic, scope, boundary and replay checks.
  • Purple-team playbook
  • Telemetry contract and deployment
  • Coverage and evidence
  • Sources and licensing

Testing and deployment

The test suite checks Sigma parsing and rule references, atomic matching, correlation order and time windows, identity boundaries, distinct counts, upstream file integrity and atomic Splunk query compilation. Synthetic fixtures cover positive sequences, benign activity and an approved-automation false positive.

Live ARTEX execution and SIEM runtime behavior have not been validated. Matches identify behavior or fingerprints, not attribution to ARTEX.

Map source telemetry to the documented normalized schema before deployment. The offline evaluator supports this pack's selection, two-stage ordered and distinct-count logic; it is not a general Sigma engine. Validate field mappings and correlation semantics in your SIEM, then baseline in audit-only mode.

Contributing and deployment readiness

Start with a reproducible fixture, expected matches and a clear telemetry mapping. Include a benign counterexample and explain tuning tradeoffs when proposing rules. Do not submit credentials, personal data or customer logs. See the deployment readiness checklist before operational use. The offline evaluator has no outbound telemetry and never executes event content.

License

Original code, documentation and original rules: MIT (LICENSE). Vendored rules: Detection Rule License 1.1 (vendor/theori/LICENSE); preserve Xint attribution, source links and license, including attribution in match output. See UPSTREAM.json for pinned Git blob identifiers and byte checksums.