Project Request: SidClaw — approval and audit layer for AI agents

## Project Request ### Project Name SidClaw ### Description SidClaw is an open-source approval and audit layer for AI agents. It intercepts agent tool calls at runtime, evaluates them against configured policies, and holds them for human review before execution. When the action is approved (or auto-approved by policy), it proceeds. Every action gets logged to a tamper-proof hash-chain audit trail. ### Category Runtime > Tool (or a new "Governance" sub-section under Runtime if appropriate) ### Homepage URL https://sidclaw.com ### Repository URL https://github.com/sidclawhq/platform ### Why does this belong in Awesome-LLMOps? LLMOps requires governing what AI agents actually do in production — not just observing after the fact. SidClaw fills the runtime governance gap between observability tools (which show you what happened) and policy documents (which say what should happen). It's the enforcement layer. Use cases common to LLMOps teams: - Block an agent from pushing to production without approval - Require human sign-off before an agent sends email to customers - Audit every tool call for compliance (FINRA 2026, EU AI Act Art. 14, NIST RMF) ### Integrations 13 framework integrations: MCP (GovernanceMCPServer), LangChain, LangGraph, CrewAI, Vercel AI SDK, OpenAI Agents SDK, AutoGen, Google ADK, and more. ### License Apache 2.0 (SDK) / FSL 1.1 (Platform) ### Logo Available at https://sidclaw.com/logo.png (PNG, square background) ### Additional Notes - 891 tests - TypeScript + Python SDKs - Live demos: https://demo.sidclaw.com - Docs: https://docs.sidclaw.com