Search

In-process MakerChecker governance. Import the pure primitives — deny-by-default authorization + separation of duties — with no server, no Postgres, and no audit chain; add the signed, offline-verifiable chain only when you want it.

Reeflex — deterministic governance gate for AI-agent actions, the Claude Code adapter (PreToolUse governance hook). Name reservation; code at github.com/Reeflex-io/reeflex

Python SDK for EvalGuard -- evaluate, red-team, and guard LLM applications with drop-in framework integrations

Multi-agent white-box and threat-modeled black-box testing harness for vibe-coded projects ／ 針對 vibe coding 專案的多代理白箱與威脅建模黑箱測試 harness

A curated collection of Claude Agent Skills designed to enhance AI workflows with specialized capabilities.

Python SDK for AgentGuard — the firewall for AI agents

Demonstrates structured logging of Agentic AI calls to MCP servers, building a Phishing Triage Assistant as an example

Interactive installer for HexStrike AI + OpenCode on Ubuntu/Kali and Windows: it scans existing tools, installs only what's missing, patches the minor bugs

Deterministic governance predicates and allow or deny decision logic for Riverbraid.

A minimal-privilege proxy for MCP tool calls

Deterministic refusal and boundary enforcement layer for Riverbraid.

Layered operator screening and session-hijack resistance for AI-agent platforms (alpha)

Native safety hooks for Claude Code — blocks destructive commands, enforces tool-call budgets, guards file reads, masks secrets, and audits every agent action. Fail-closed C++ daemon with extensible rule engine.

Open-source MCP server that lets AI agents call APIs without ever seeing the credentials. Policy-enforced secret injection into HTTP calls and subprocesses, with encrypted audit trail.

Reproducible benchmark of denial-of-wallet attacks and defenses for LLM agents

Pure-function policy matrix evaluator for AI coding agents (repo x capability x context -> deny/require_approval/auto_allow).

Full-featured agent toolkit for the DashClaw platform. 178+ methods across 30+ categories for governance and orchestration.

证据驱动的 FOFA 资产测绘智能体：自然语言侦察、AI 反思、CLI / MCP / Skill / REST API，以及经人工审批的 Nuclei 扫描。

Tamper-evident, hash-chained Runtime Records for AI agents: the TypeScript recorder. Chain-format compatible with the Python halo-record package.

v0.6 signing-time gate for Solana: decode any base64 tx offline, classify 37 danger primitives + Native Stake + bounded ATA/Memo recognition + a 14-program DeFi/NFT clear-signing registry, resolve ALTs and clear-sign Squads proposals with --rpc, optionall