Open-source AI Security Operations Center: alert fusion, LLM-agent triage, MITRE ATT&CK investigation, and a replayable decision ledger for every agent step. Self-hostable, runs with no API keys, MIT licensed. Ships an MCP server for Claude, Cursor and Continue.
Detection Importers
Tools for pulling detection content from upstream open-source projects and converting it into AiSOC's native YAML schema. Every imported rule carries a populated provenance block so the source is always traceable.
Layout
| Importer | Upstream | Output | Status |
|---|---|---|---|
sigma_importer.py |
SigmaHQ/sigma | detections/sigma-imports/<category>/ |
working |
car_importer.py |
mitre-attack/car | detections/car-imports/ |
working |
chronicle_importer.py |
chronicle/detection-rules | detections/chronicle-imports/ |
scaffolded |
splunk_importer.py |
splunk/security_content | detections/splunk-imports/ |
scaffolded |
import_orchestrator.py |
(orchestrator) | runs every importer in turn | working |
Provenance schema
Every imported rule carries a top-level provenance block:
provenance:
source: SigmaHQ/sigma # short upstream identifier
source_id: 5e957c0a-... # upstream rule id (UUID for Sigma, CAR-YYYY-MM-NNN for CAR, etc.)
source_commit: a1b2c3d # short SHA of the upstream commit we imported from
license: DRL-1.1 # upstream license SPDX-ish identifier
license_url: https://github.com/SigmaHQ/Detection-Rule-License
imported_at: 2026-05-04 # YYYY-MM-DD when the importer wrote this file
imported_by: sigma_importer # which importer produced this file
upstream_path: rules/cloud/aws/aws_root_login.yml # original path inside the upstream repo
This block is required for every rule under detections/sigma-imports/, detections/car-imports/, detections/splunk-imports/, and detections/chronicle-imports/. The validator (scripts/validate_detections.py) enforces it.
Native rules under detections/cloud/, detections/identity/, etc. carry provenance.source: native and do not require a source_commit.
Quality tiers
| Tier | Path | Validation | Fixture required | Visibility |
|---|---|---|---|---|
| Native | detections/<category>/ |
full schema + fixture replay | yes | tier: stable |
| Imported (verified) | detections/<source>-imports/<category>/ |
schema + parse + MITRE mapping | no | tier: imported |
| Imported (quarantine) | detections/<source>-imports/_quarantine/ |
parse only, enabled: false |
no | tier: quarantine, hidden by default |
| Community | detections/community/ |
schema + fixture replay | yes | tier: community |
Running an importer
Each importer is idempotent and pinned to a specific upstream commit. To upgrade the corpus:
# 1. update the SIGMA_COMMIT, CAR_COMMIT, etc. constants in tools/detection_import/import_orchestrator.py
# 2. run the orchestrator (from the repo root):
python3 -m tools.detection_import.import_orchestrator
# 3. validate
python3 scripts/validate_detections.py
# 4. rebuild the marketplace index
pnpm marketplace:build && pnpm marketplace:check
The orchestrator clones each upstream repo into .import-cache/ (gitignored), reads from a pinned commit, and writes converted rules to the appropriate detections/<source>-imports/ directory.
Why pinned commits?
Detection rules drift. SigmaHQ moves a rule from experimental/ into stable/, the CAR project renames a model, Splunk security_content reshapes its YAML. Pinning to a commit means our shipped corpus is reproducible: anyone running the orchestrator against the same commit gets the same rule files. Bumping is a deliberate PR.
Adding a new importer
- Create
tools/detection_import/<name>_importer.pywith adef import_rules(commit: str, *, output_root: Path | None = None) -> list[ImportedRule]entrypoint. - Wire it into
tools/detection_import/import_orchestrator.py. - Document the upstream license in
.github/LICENSES.md. - Update
detections/README.mdwith the new tier.