--- id: "nr-long-running-agent-work-is-becoming-durable-infrastructure-not-a-chat-3b8de2" slug: "long-running-agent-work-is-becoming-durable-infrastructure-not-a-chat-3b8de2" type: "raw_signal" title: "Long-running agent work is becoming durable infrastructure, not a chat session" description: "Claude Code added MCP Tasks for durable long-running jobs, and Vercel's durable-harness resumes interrupted coding-agent work." observed_at: "2026-10-24" record_date: "2026-10-24" date_kind: "observed_at" novelty: "notable" source_platform: "newruntime.com" topics: ["claude","claude-code","mcp"] entities: ["Anthropic","Claude Code","Model Context Protocol","MCP Tasks","MCP Elicitations","Vercel","durable-harness","Pi","OpenCode","Codex"] related_patterns: ["agent-protocols-become-interoperability-layer","goal-scoped-loops-replace-manual-continuation","agent-ready-software-exposes-capabilities"] source_url: "https://code.claude.com/docs/en/mcp" source_urls: ["https://code.claude.com/docs/en/mcp","https://vercel.com/blog/making-every-harness-durable","https://www.notion.com/blog/notion-ai-more-for-less","https://pypi.org/project/managed-deepagents/0.9.0","https://www.peterp.me/articles/what-actually-happens-when-you-message-grok-bot","https://cloud.google.com/blog/topics/ai-infrastructure/introducing-agent-compute-engine-and-expanded-agent-governance"] why_it_matters: "Claude Code added MCP Tasks for durable long-running jobs, and Vercel's durable-harness resumes interrupted coding-agent work. Notion's agents now run for up to 24 hours, LangChain added managed schedules, and Grok Bot runs on queues, workers and retries. Google launched microVM sandboxes built for agent workloads. Builders should treat retries, resumption and scheduling as core requirements for agents and stop expecting them to finish within a single conversation turn." verification_level: "source-linked" signal_type: "synthesis" synthesis_of: [{"slug":"claude-code-cli-gains-mcp-tasks-for-durable-1711e7","title":"Claude Code CLI gains MCP Tasks for durable long-running jobs and Elicitations for mid-task user input.","url":"https://newruntime.com/posts/claude-code-cli-gains-mcp-tasks-for-durable-1711e7/"},{"slug":"making-every-harness-durable-analysis-9f2d73","title":"Making every harness durable","url":"https://newruntime.com/posts/making-every-harness-durable-analysis-9f2d73/"},{"slug":"agents-rebuilt-to-do-more-for-less-analysis-76f6d4","title":"Agents, rebuilt to do more for less","url":"https://newruntime.com/posts/agents-rebuilt-to-do-more-for-less-analysis-76f6d4/"},{"slug":"managed-deepagents-0-9-0-released-on-pypi-a59d19","title":"managed-deepagents 0.9.0 released on PyPI with managed schedules and per-run agent factories","url":"https://newruntime.com/posts/managed-deepagents-0-9-0-released-on-pypi-a59d19/"},{"slug":"stanford-cs146s-lecture-by-lee-robinson-explains-how-dc4775","title":"Stanford CS146S lecture by Lee Robinson explains how Grok Bot's always-on agent works","url":"https://newruntime.com/posts/stanford-cs146s-lecture-by-lee-robinson-explains-how-dc4775/"},{"slug":"google-cloud-launches-agent-compute-engine-for-sandboxed-229085","title":"Google Cloud launches Agent Compute Engine for sandboxed agent runtimes","url":"https://newruntime.com/posts/google-cloud-launches-agent-compute-engine-for-sandboxed-229085/"}] schema_version: "newruntime-agent-readable-v0.2" stable_id: "signal:long-running-agent-work-is-becoming-durable-infrastructure-not-a-chat-3b8de2" retrieval_nugget: "Claude Code added MCP Tasks for durable long-running jobs, and Vercel's durable-harness resumes interrupted coding-agent work. Notion's agents now run for up to 24 hours, LangChain added managed schedules, and Grok Bot runs on queues, workers and retries." status: "published" visuals: [] editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]} routes: {"html":"https://newruntime.com/signals/long-running-agent-work-is-becoming-durable-infrastructure-not-a-chat-3b8de2/","markdown":"https://newruntime.com/signals/long-running-agent-work-is-becoming-durable-infrastructure-not-a-chat-3b8de2.md","json":"https://newruntime.com/signals/long-running-agent-work-is-becoming-durable-infrastructure-not-a-chat-3b8de2.json"} --- # Long-running agent work is becoming durable infrastructure, not a chat session ## Retrieval answer Claude Code added MCP Tasks for durable long-running jobs, and Vercel's durable-harness resumes interrupted coding-agent work. Notion's agents now run for up to 24 hours, LangChain added managed schedules, and Grok Bot runs on queues, workers and retries. Claude Code added MCP Tasks for durable long-running jobs, and Vercel's durable-harness resumes interrupted coding-agent work. Notion's agents now run for up to 24 hours, LangChain added managed schedules, and Grok Bot runs on queues, workers and retries. Google launched microVM sandboxes built for agent workloads. Builders should treat retries, resumption and scheduling as core requirements for agents and stop expecting them to finish within a single conversation turn. From the Field Notes: - [Claude Code CLI gains MCP Tasks for durable long-running jobs and Elicitations for mid-task user input.](https://newruntime.com/posts/claude-code-cli-gains-mcp-tasks-for-durable-1711e7/) (2026-10-24) - [Making every harness durable](https://newruntime.com/posts/making-every-harness-durable-analysis-9f2d73/) (2026-10-22) - [Agents, rebuilt to do more for less](https://newruntime.com/posts/agents-rebuilt-to-do-more-for-less-analysis-76f6d4/) (2026-10-22) - [managed-deepagents 0.9.0 released on PyPI with managed schedules and per-run agent factories](https://newruntime.com/posts/managed-deepagents-0-9-0-released-on-pypi-a59d19/) (2026-10-24) - [Stanford CS146S lecture by Lee Robinson explains how Grok Bot's always-on agent works](https://newruntime.com/posts/stanford-cs146s-lecture-by-lee-robinson-explains-how-dc4775/) (2026-10-24) - [Google Cloud launches Agent Compute Engine for sandboxed agent runtimes](https://newruntime.com/posts/google-cloud-launches-agent-compute-engine-for-sandboxed-229085/) (2026-10-23) Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json ## Continue in the graph Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`. - topic hub: [Claude](https://newruntime.com/topics/claude.md): Explore the claude topic hub. - topic hub: [Claude Code](https://newruntime.com/topics/claude-code.md): Explore the claude-code topic hub. - topic hub: [Mcp](https://newruntime.com/topics/mcp.md): Explore the mcp topic hub. - collection: [Raw Signals](https://newruntime.com/signals.json): Open every source-linked raw signal. - collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph. - whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)