kars

Pair with a kars cluster and offload heavy tasks to governed cloud sandboxes with GPU / foundation-model inference / Azure AI services, or communicate with other agents over end-to-end encrypted AgentMesh. Triggers on natural-language intents like "offload to the cloud", "run this on Azure", "ask my cluster to…", "send a message to agent X", "who is on the mesh", "check my inbox", "is my offload done".

Behavioral governance for OpenClaw agents via AGT — tool-level policy, inter-agent trust, audit logging.

Spawn secure isolated sub-agent sandboxes, delegate tasks via AGT mesh, receive results, and destroy sub-agents. Uses the kars_spawn, kars_mesh_send, kars_mesh_inbox, and kars_spawn_destroy tools.

Query and inspect Foundry prompt agents and invoke Foundry tools via the Responses API. OpenClaw is the orchestrator — Foundry provides managed AI services.

Python code execution via Azure AI Foundry Responses API with code_interpreter tool. Data analysis, charts, and math in a managed sandbox.

Manage persistent conversations via Foundry Conversations API. Create conversations, add messages, and maintain history across sessions.

Query model deployments, connections, and indexes in the Foundry project. Discover available models and infrastructure.

Evaluate agent quality using Foundry OpenAI Evals API. Create evaluations, run them against models, and analyze results.

Knowledge retrieval (RAG) via Foundry file_search and azure_ai_search tools. Agentic retrieval with citations — uses Responses API.

Persistent long-term memory via Foundry Memory Store APIs. User preferences and chat summaries survive pod restarts — no Foundry hosted agent needed.

Real-time web search via Azure AI Foundry Responses API with bing_grounding tool. Get current information with citations — no egress policy needed.