arXiv:2611.21978v1 Announce Type: new Abstract: Large language models are moving from isolated assistants toward autonomous agents that can plan, use tools, and coordinate with others, motivating an emerging Agentic Web of independently developed agents and services. Realizing this vision requires agent behavior to move across creators, hosts, and runtimes without coupling publication to a particular execution environment or transferring authority over user resources. We introduce WebAgents, which unbundles these roles through a signed agent.pkg that carries portable workflows, skills, schemas, and resource requirements. An OpenClaw-based agent-native browser/runtime verifies the package, obtains authorization, and binds it at run time to user-controlled models, tools, permissions, and budgets, while hosts distribute the public program and may retain private services server-side. We demonstrate this contract with three unrelated workflows executed unchanged under local and API-backed models, including literature review with live services and a website-published travel agent using semantic tools and user-selected personal resources. In a controlled suite, WebAgents explicitly blocks all 716 injected contract violations while accepting all 295 matched benign controls, with median automated pre-execution processing below 0.5 s for all three packages. We conclude by outlining a path toward NANDA-based discovery and interoperable package ecosystems while retaining execution authority in the user's runtime.
WebAgents: Unbundling Agent Creation, Hosting, and Execution for an Agent-Native Web
A paper posted to arXiv (2611.21978v1) introduces WebAgents, a system that unbundles agent creation, hosting and execution through a signed agent.pkg carrying portable workflows, skills, schemas and resource requirements, verified at run time by an OpenClaw-based agent-native browser/runtime that binds the package to user-controlled models, tools, permissions and budgets. In a controlled suite, WebAgents blocked all 716 injected contract violations while accepting all 295 matched benign controls, with median automated pre-execution processing below 0.5 seconds for all three packages, and three unrelated workflows ran unchanged under both local and API-backed models. The authors outline a path toward NANDA-based discovery and interoperable package ecosystems while keeping execution authority in the user's runtime.
Verify signed agent packages against declared permissions and resource requirements in the user's runtime before execution; the source blocked all 716 injected violations with sub-0.5-second median checks.
more in playbook: Securing agents and AI toolsRun your AI side-project on zahid.host
EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.