Distributed agent coordination platform where agents live, connect, coordinate, execute, and evolve together — like an organization.

An educational AI coding agent CLI

Open-source email API for applications and AI agents. Managed hosting at e2a.dev, or self-host with Docker.

A traceable game localization pipeline with translation memory, model-assisted workflows, QA gates, human review, artifact builds, and multi-target publishing.

A LangGraph template for building AI agents with human-in-the-loop, conditional routing, observability (Langfuse), and a Reflex UI demonstrated through a travel planner use case.

Score the trustworthiness of outputs from any LLM in real-time

Semantic Flow Language (SFL) is a framework that aligns human intent, AI reasoning, and executable logic. It ensures bidirectional synchronization between meaning and code, allowing for reliable, transparent, and verifiable AI-driven development across various environments.

Governance beneath the model. Custody before trust. Open for audit. Constitutional Grammar for Multi-Model AI Federations, Firmware Specification • Zero-Touch Alignment • Public Release v1.0

12-week Hybrid Agentic System Pilot for a fintech. THRIVE capstone winner. Full documentation, deliverables, DB schema & live demo.

Plug-and-play Human-In-The-Loop integration of agentic workflows

MCP governance proxy for AI agents — enforce rate limits, access control, human approval, and audit logging at the protocol level.

Manual Verification Protection + AI Selector Compatible with OpenClaw . This is a complete TypeScript plugin project—a true implementation of the `before_tool_call` hook that can intercept tool calls, follow OpenClaw’s built-in approval workflow, and features an approval window mechanism.

AWS-first, human-in-the-loop FinOps platform — turns cloud cost insight into approval-gated action using Amazon Bedrock Agents, FastAPI, and Next.js 15.

Governed AI engineering control plane: federated workload identity, delegated least privilege, pre-dispatch MCP/API authorization, human approval, revocation, and cryptographically signed authority receipts with tamper/replay/rollback detection and independent verification. AI proposes. Humans decide.

Security & runtime checklist for moving LLM agents from demo to production: HITL gates, idempotency, prompt injection boundaries, tool risk classification (CC BY 4.0).

Policy enforcement and execution infrastructure for AI agents. Merchval turns business rules into controlled, auditable actions using OPA, trusted evidence, approval workflows, and fail-closed decisions—helping teams deploy agentic AI safely across e-commerce, insurance, banking, and other high-stakes operations.

Fraud investigation tutorial across 9 phases — same 6 cases, progressively adding LangGraph, tools, HITL, multi-agent coordination, and LangSmith observability

Experimental source-available AGI research architecture: reality-coupled cognition, bounded recovery, lifelong learning, representation synthesis, structural transfer, shadow strategy evaluation, persistent memory, and human-governed execution. AI proposes. Humans decide.

AI curator agents bid on artworks; every commitment of funds passes through a human gesture. POC-001/002/003/004 verifiable on Base Sepolia. Agent governance documented. Handmade, solo. INPI DSO2026016080 · DSO2026023753 · DSO2026025380

DSH 插件：把权限审批和模型提问改成大白话；审批按风险分级并列出会改动的文件路径。Plain-language approvals and question prompts for DeepSeek Harness.