10 projects
posturekit
Ingest large, messy data-security exports (.xlsx/.csv) into DuckDB and score them for risk. A model-agnostic DSPM kernel — bring your own data and risk matrix.
slidecast
Turn a list of HTML slides + narration into a narrated MP4. Headless-browser screenshots, pluggable text-to-speech, ffmpeg stitching. Bring your own slide design and voice.
quizforge
Generate a deep, mixed-format question bank from source material and grade it — deterministic where it can, LLM where it must. Bring your own chat model.
aisoc
An AI SOC kernel: an event-sourced, multi-agent SOC where a role-based team of LLM agents works a case over a shared event bus — with case memory, human-in-the-loop gating, and a replayable audit log. Bring your own LLM, alert source, and tools.
refutescan
A two-LLM, refute-first agentic source-code vulnerability scanner — wide-net navigate, skeptical refute, jailed in docker. Bring your own LLM.
attestq
Answer security questionnaires and compliance attestations from your evidence — a model-agnostic RAG kernel.
langchain-failover
Multi-model orchestration for LangChain chat models: primary/secondary failover (resilience) and tier-split gather/compose (cost/latency), tool-calling preserved across both.
detflow
Draft detections from plain English and review them like a senior detection engineer — Sigma and Cortex XQL, offline-safe, model-agnostic.
domainflow
Discover, monitor, score, and cluster brand-impersonation domains — a layered, pip-extras toolkit for the lookalike-domain lifecycle, from typo-squat generation to actor-campaign clustering.
iocflow
Extract, enrich, assess, hunt, block, and investigate indicators of compromise (IOCs) — a layered, pip-extras toolkit for the full IOC lifecycle, from unstructured text to feed-driven triggers and a human-in-the-loop multi-agent response.
