tool-governance

让 Agent 通过对话直接操作你的商城、SaaS、CRM 或 ERP 后台，同时保留原有权限、审批与审计。Self-hosted hub for agents operating existing business systems safely.

🦾 The Agent Runtime Where Every Thought Is Visible — 7 paradigms, 10 harness shells, zero black boxes.

The approval and accountability layer for agentic AI. Identity → Policy → Approval → Trace. Try: npx sidclaw-demo

Self-hosted model routing gateway and agent control plane for OpenAI, Anthropic, Gemini and A2A agents behind one OpenAI-compatible API, protocol-first. MCP server security, governed tools, prompt enhancement, virtual-key budgets, cache, observability, audit chain and agent governance. Zero-key local first run; public preview.

Agent control plane and runtime access platform for enterprise systems — govern business APIs as Tools, compose Agents and Workflows, expose runtimes through AAP, and trace executions end to end. https://act-weave.grok.me

System of record for AI tool risk: inventory, policy enforcement, approvals, and audit-ready evidence.

Run a Neura Relay Action Card and receive a governed Decision Receipt before execution.

Harness engine for AI Agents. From demo to production.

Open Agent Capability Contract for Agent-to-Business capability declaration

ClawFlow: A lightweight Agent Runtime / AgentOS Kernel for next-generation personal AI agents

Verifiable coding-agent runtime for Python-based AI agents with evaluation, browser automation, and tool governance.

Deterministic MCP runtime for safe tool execution, policy gates, contracts, and local AI workflow governance.

ForceField Python SDK -- AI security in 3 lines of code. Prompt injection detection, PII redaction, security evals, tool governance. GitHub Action, pre-commit hook, Homebrew, VS Code extension.

Fleet-analyze a directory of MCP Tool Cards. Counts by side-effect class, PII/secrets exposure, reversibility, human-approval-required; surfaces approval gaps (destructive tools missing human_approval_required: true). Library + CLI.

Deterministic pre-execution gate for one credentialed tool request: explicit policy, allow/deny decision, and inspectable decision artifact....verifiable receipts for multi-agent systems.

Paid remote MCP for schema drift checks, tool-schema approvals, compatibility receipts, breaking-change explanations, and release audit logs.

基于 Open WebUI 与 Claude Agent SDK 的多用户研发协作工作台：自定义模型、权限隔离、检索编码协作、独立验收与可审计交付

Shared tool discovery and execution for AI clients and applications across MCP, A2A, and internal tools—intent-ranked Top-K retrieval with centrally managed provider and tool changes

Python FastAPI service for evaluating MCP server and tool policies, trust posture, destructive-action controls, and operator-facing review workflows.