Skip to main content
Avatar for Stefano Rizzello from gravatar.com

Stefano Rizzello

Username    Rinkia
Date joined   Joined

12 projects

bastiongateway

Last released

MCP security gateway: an inline proxy that scans tools/list for poisoned tools, enforces a tool allow/deny policy, blocks injected tool results, and logs every call. The runtime-enforcement leg of the bastion family.

bastioncorpus

Last released

The shared prompt-injection corpus for the bastion trilogy: one canonical, versioned dataset that agentbastion (prevent), bastionprobe (attack), and bastiontrace (investigate) all pull from.

bastiontrace

Last released

Forensics for injected AI agents: read a tool-call trace, locate the prompt injection, and map its blast radius. Investigate side of the bastion trilogy (agentbastion + bastionprobe).

bastionfuse

Last released

Kill switch and tripwires for AI agents: honeytokens, decoy files, decoy MCP tools and action budgets trip a sticky fuse that stops the agent, whether or not any detector saw the injection. Claude Code hook + Python library. The CONTAIN leg of the bastion family.

bastionsupply

Last released

MCP supply-chain security scanner: detect tool-poisoning, shadowing, hidden unicode, secret solicitation, and rug-pull drift in MCP servers before you install them.

bastionmesh

Last released

Runtime security gateway between A2A agents: scans peer replies for prompt injection, catches injected payloads relayed between agents, pins and rewrites agent cards, and caps delegation depth, cycles and fan-out.

agentbastion

Last released

A firewall for AI agents: block prompt injection, guard tool calls, redact PII/secrets.

bastionprobe

Last released

Pentest for AI agents: fire indirect prompt-injection payloads at your agent and report which land. Offensive twin of agentbastion.

bastionskill

Last released

Skill-poisoning scanner: detect malicious bundled code (network egress, secret theft, hook-install persistence, destructive commands) in agent skills before you install them — the code-layer that a plain grepper and a prompt-scanner miss.

bastionmemory

Last released

Agent memory-poisoning scanner (OWASP Agentic ASI06): detect injected directives disguised as facts in what an AI agent remembers, and diff memory against a baseline to catch poison that appears over time.

modscan

Last released

Scan a codebase, generate everything needed to write plugins and mods for it

monohunter

Last released

Find single long-period (mono-)transits in public TESS light curves that periodic pipelines under-find.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page