# Wisdoverse Forge: A Self-Hosted Governance Platform for Enterprise AI Workflows > Wisdoverse Forge is a self-hosted AI workbench built with Rust that provides auditable task, run log, and evidence chain management. It supports three agent runtimes (container, host CLI, and API) and integrates the Temporal workflow engine with a strict Feature-Sliced Design frontend architecture. - 板块: [Openclaw Llm](https://www.zingnex.cn/en/forum/board/openclaw-llm) - 发布时间: 2026-05-28T19:15:20.000Z - 最近活动: 2026-05-28T19:19:53.052Z - 热度: 161.9 - 关键词: AI workbench, Rust, self-hosted, Temporal, governance, audit, agent runtime, FSD, enterprise AI - 页面链接: https://www.zingnex.cn/en/forum/thread/wisdoverse-forge-ai - Canonical: https://www.zingnex.cn/forum/thread/wisdoverse-forge-ai - Markdown 来源: floors_fallback --- ## Introduction / Main Floor: Wisdoverse Forge: A Self-Hosted Governance Platform for Enterprise AI Workflows Wisdoverse Forge is a self-hosted AI workbench built with Rust that provides auditable task, run log, and evidence chain management. It supports three agent runtimes (container, host CLI, and API) and integrates the Temporal workflow engine with a strict Feature-Sliced Design frontend architecture. ## Original Author and Source - **Original Author/Maintainer**: Wisdoverse Team - **Source Platform**: GitHub - **Original Title**: Wisdoverse-Forge - **Original Link**: - **Publication Date**: 2026-05-28 ## Project Overview Wisdoverse Forge is a self-hosted AI workbench designed for enterprise scenarios. Its core goal is to transform team collaboration into auditable tasks, run logs, and evidence chains. Unlike most AI tools on the market that focus on rapid prototype validation, Forge has focused on Governance, Auditability, and Security from the very beginning, aiming to provide a reliable AI agent execution infrastructure for enterprise environments requiring compliance control. The project uses Rust as the core backend language, paired with React/Vite/Three.js to build the frontend interface. The overall architecture emphasizes strict Domain-Driven Design (DDD) and Feature-Sliced Design (FSD) layering. This technology selection reflects the team's emphasis on type safety, performance, and enterprise-level maintainability. ## Three Agent Runtimes: Balance Between Flexibility and Security Wisdoverse Forge's most distinctive feature lies in its support for three distinct agent runtime modes, each specifically designed for different usage scenarios and security requirements: ## Container Runtime This is a platform-hosted Docker container environment, running containerized CLI tools such as Claude, Codex, Gemini, or OpenCode, and equipped with a dedicated Sidecar agent. The container runtime provides a fully isolated execution environment for each agent, suitable for handling sensitive data or tasks requiring strict resource limits. The platform ensures container startup speed and runtime security through Docker security policies and Warm Pool technology. ## Host CLI Runtime This mode allows operators to register CLI tools from their local machines onto the platform. Agents connect to the control plane via Sidecar and NATS message bus, supporting idempotent registration (Idempotency-Key) and atomic audit event logging. The host CLI mode is suitable for developers to use familiar toolchains in their local environments while maintaining integration with the platform's governance capabilities. ## API Runtime This is the most lightweight mode, directly calling APIs from cloud service providers like Anthropic, OpenAI, and Google, with no containers or local shells required. The API runtime is suitable for rapid prototype validation or handling lightweight tasks that do not involve sensitive data, and also provides a feasible access solution for resource-constrained environments. These three runtimes ensure type safety through database-level CHECK constraints, preventing runtime configuration inconsistencies. ## Rust Backend: Foundation of Performance and Security Forge's backend uses a modular Rust workspace design, consisting of multiple crates focused on single responsibilities: - **core**: Shared domain types, error handling, and runtime type definitions - **db**: SQLx-based database connection pool and migration management - **auth**: JWT + Argon2 authentication system, including middleware implementation - **api**: REST API, WebSocket gateway, and MCP (Model Context Protocol) bridge built with the Axum framework - **orchestrator**: Temporal-based workflow orchestration engine - **platform**: Docker integration, security policies, and container pool management - **llm**: Multi-provider LLM gateway, unified interface for calling different models This layered architecture enforces the dependency direction of Route → Service → Domain → Repository through test cases, ensuring that business logic does not leak into the infrastructure layer.