Persistent memory layer for LLM agents. Every write goes through an anti-confabulation admission gate — does the source actually entail the fact? — and every read returns provenance. MCP-native: plugs into Claude Code, Cursor, Cline, Zed.
Autonomous cybersecurity platform: 50+ phase web-audit pipeline, 44 tools, 25 orchestrator modes, WSL Kali integration, bug-bounty mode. Its 5-bubble blind-validation pipeline found and fixed 112 real bugs across 3 blind audit rounds.
MCP server for adversarial code review: 3 independent workers (falsification, caller-verification, counterexample) that attack a claim before it ships. It caught production bugs its author missed — that's the point.
More projects
Agentic multi-provider desktop cockpit (Tauri + React): one workspace to drive AI across your machine, with system awareness and trust rendered on screen.
Autonomous trading suite: multi-LLM orchestrator, deterministic backtest engine, regime-driven strategy selection, full audit trail. An engineering project — no returns claimed, not a signal service.
Post-quantum crypto-agility audit for the Italian market: finds where you use cryptography, your "harvest-now-decrypt-later" exposure, and a migration plan to NIST PQC. NIS2 / DORA / AgID reports. Never exploits, no telemetry.
github.com/aureliocpr-ctrl/pqc-audit-italia ↗Runtime epistemic profiler for AI agents — tags every claim as grounded, ungrounded or unverifiable. The trust thesis as a standalone tool.
github.com/aureliocpr-ctrl/groundprobe ↗Cross-LLM CLI orchestrator: drive Claude, Gemini and Codex from one control plane over a shared memory bus. The workshop behind the products.
Benchmark harness comparing frontier models on real coding tasks — measured, reproducible.
github.com/aureliocpr-ctrl/sakana-bench ↗Digital intercom for short-term rentals: legal in-person check-in, guest flows. Proof I ship beyond AI, too.
github.com/aureliocpr-ctrl/citofono ↗A small Italian town where AI characters live their own lives — a multi-agent social simulation. Early-stage, filed honestly as a sandbox.
From a one-hour consultation to a complete build: AI adoption, custom software, assistants, automation, reliability engineering, security audits. Everything verifiable is on GitHub; the rest I'll show you live. Remote worldwide, based near Milan.