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.
Python agentic-aiai-securityalert-triageblue-teamcybersecuritydetection-engineeringincident-responselanggraphllm-agentsmcpmitre-attackmodel-context-protocolpurple-teamsecurity-operationsself-hostedsiemsoarsocthreat-detectionthreat-intelligence
"""Closure policy and the kill switch.
Its own package rather than a module under ``app.policy``, deliberately.
``app/policy/__init__.py`` re-exports everything in ``guardrails``, so
importing any sibling from there executes that re-export and makes a module
with no caller look reachable to ``scripts/check_module_reachability.py``.
Placing this here keeps that gate's verdict on ``guardrails`` honest.
"""
from .policy import (
POOL_READY,
POOL_UNAVAILABLE,
POOL_UNCONFIGURED,
ClosureDecision,
decide,
kill_switch_engaged,
reset_cache,
resolve_pool,
shared_pool,
)
# `authorization` is deliberately NOT re-exported here. It imports
# `app.agents.dispositions`, and `app.agents` reaches back into this package,
# so eagerly importing it from the package root would be a cycle. Callers
# import `app.closure.authorization` directly, which is also the honest
# spelling for a chokepoint: it should be obvious in a diff when a module
# starts depending on it.
__all__ = [
"POOL_READY",
"POOL_UNAVAILABLE",
"POOL_UNCONFIGURED",
"ClosureDecision",
"decide",
"kill_switch_engaged",
"reset_cache",
"resolve_pool",
"shared_pool",
]