Havuz

Claude Code-style declarative permission rules for DeepSeek Harness: ordered allow/deny/ask rules with tool-name, argument (glob/regex), and

Parent-operated starter kit for designing a private, child-facing Hermes Agent profile

Open-Source Agentic Harness for Reliable Task Completion, even on small local models. 💻🧠🤖

Recovering verbatim private reasoning from Claude extended-thinking signatures, with a local byte-exact verifier.

Verify what AI coding agents can actually read, write, execute and reach using synthetic canaries.

Real-time web dashboard for AI-agent behavior monitoring, built on Perplexity's Numbat. Zero-dependency Python.

不死鸟 Phoenix — Hermes Agent 插件：路由分档/风险防线/自愈/存档点提醒/审批策略自适应，官方钩子接入，不改 Hermes 核心代码

Six skills tell coding agents how to work. Two guards enforce the parts that should not depend on judgment.

Streaming privacy for LLM apps: pii-leak-benchmark (does your gateway leak PII in streams?), chunk-invariance (one-line test for split-bound

Bilingual graduate textbook on LLM post-training: SFT, RLHF, DPO, GRPO, RLVR, reasoning, agents, systems, evaluation, and safety.

1,017 system prompts from real AI products, audited span by span against the eight AISPA assurance dimensions

Open-source runtime security and execution control for autonomous AI agents. ALLOW, BLOCK, ESCALATE at the execution boundary with auditable

Vault, don't delete — hallucination governance for long-memory AI companions. 梦,归本人。

Checks every AI agent action against your company's real rules before it runs. Git-native, PR-reviewed.

The gate between an AI agent and real money. Runtime authorization for financial AI agents: structured intent, deterministic policy, human a

Agent Execution Partnership AEE is an open-source control plane that ensures every AI agent action is authorized before it runs, observable

Open-source AI security verification for model artifacts, live endpoints, MCP servers, and recorded agent traces. Reproducible evidence for

A modular threat detection library for AI and LLM pipelines. It provides language-agnostic guardrails through a configurable pipeline of loa

Overnight experiment: can an AI agent make money, and can you trust what it reports? The agent produces claims; a separate verifier it canno

Local-first verification layer for AI agents: evidence, provenance, policy gates, human approval, and auditable Trust Receipts before a clai

Turn Android Termux into an autonomous AI agent and security orchestrator. Powered by a local ReAct framework with 50 system tools and MCP s

Context bomb strings

Python SDK for LLM guardrails with safety classification, PII detection, prompt injection defense, and grounding checks. Protect AI applicat

Grade every agent, scraper and model in a claim's chain — provenance, trust scoring and audit evidence for LLM pipelines

A universal, industry-neutral taxonomy of cognitive core skills (perception, memory, reasoning, planning, action, verification, learning, go

TypeScript SDK for IdentArk's AgentGateway: AI agents get capabilities, not credentials, with human approval, budgets and audit on every act

A Socratic supervision layer for AI coding agents.

OpenGuardrailsBench (OGB): an open, reproducible benchmark for AI guardrails across safety, security, privacy and policy

Noisegate: a differential privacy gateway that lets an untrusted LLM agent query sensitive data over MCP (Model Context Protocol), with a fo

Guardrails for AI coding agents — block catastrophic tool calls before they run. Semantic, agent-neutral, near-zero false positives.

Diagnosing policy grounding in LLM judges via causal erasure. Companion code to the AIES 2026 submission READ.

CatchBench: a benchmark for auditing agent failures across the full lifecycle.

Kernel-enforced authority and spend platform for AI agents

Subagent Verification for Claude AI Code Networks 2026

Automated Proof-of-Carrying Change Management for AIOps 2026

A small, vetted, self-evolving harness for Claude Code — curated, not dumped. Skills, safety guards, and a self-improvement loop.

Single-agent multi-modal system for insurance damage claim verification. #1 globally, HackerRank Orchestrate June 2026 (15,295 registrants).

🔥 CogitoAgent — An autonomous AI agent, cloud-driven but locally executed. It continuously thinks, automatically analyzes, and integrates pr

Methodology and deterministic evaluation tools for iterative improvement of skills, repositories and agent workflows, with independent revie

sofagent is an open-source FDE Harness for AI coding agents — on entry, write your business judgment into files (workflows, ontology, AI nod

Mechanical Governance for LLM Decisions — model-agnostic governance regimes (R1/R2/R3), hard gates, entropy commit-reveal and governance met

An AI that administers Linux through typed, approval-gated, Ed25519-audited actions. The model selects the action; the daemon builds the com

GateMem: a benchmark and evaluation toolkit for memory governance in multi-principal shared-memory LLM agents.

Audit any agent decision across its past, present, and future, on one typed graph.

Alignment-research scaffold (autoresearch-style) for LLM guardrails: search over a single policy.md surface

AI Initiative Ledger — local spend meter and budget gate for AI tasks. Meters every vendor leg (LLM, speech, telephony, search, tools), hard

Code-First AI Compliance. The open-source regulatory engine for the EU AI Act. Automate your risk assessments directly within your CI/CD pip

High-fidelity Claude Fable 5 (Mythos) environment emulation and automated multi-agent jailbreak (Pack Hunt) research laboratory.

SHACKLE — a governance protocol and policy-decision daemon for autonomous AI agents. Enforces guardrails, budget/loop limits, and policy con

Session-layer circuit breaker for LLM & agent traffic — a transparent streaming proxy that meters every session and kills the one that goes