{ "schema_version": "newruntime-topics-index-v0.1", "type": "topic_hub_index", "count": 33, "routes": { "html": "https://newruntime.com/topics/", "json": "https://newruntime.com/topics.json" }, "topics": [ { "slug": "coding-agents", "label": "Coding agents", "description": "A living map of coding agents as they move from autocomplete into supervised software work, review loops, environments, and operating cost.", "status": "featured", "last_updated": "2026-07-22", "counts": { "total": 49, "signals": 39, "patterns": 1, "posts": 6, "atlas": 3, "sources": 81 }, "routes": { "html": "https://newruntime.com/topics/coding-agents/", "markdown": "https://newruntime.com/topics/coding-agents.md", "json": "https://newruntime.com/topics/coding-agents.json" }, "answer": [ "Coding agents are becoming operational workers inside software teams, not just better code completion.", "The durable bottlenecks are review bandwidth, environment setup, context quality, permissions, and cost per completed task.", "The useful question is no longer whether an agent can write code, but whether the surrounding harness can verify and land work repeatedly." ] }, { "slug": "context-engineering", "label": "Context engineering", "description": "How teams package memory, rules, traces, tools, and task state so agents can work without restarting from raw chat context.", "status": "featured", "last_updated": "2026-07-21", "counts": { "total": 16, "signals": 9, "patterns": 1, "posts": 3, "atlas": 3, "sources": 27 }, "routes": { "html": "https://newruntime.com/topics/context-engineering/", "markdown": "https://newruntime.com/topics/context-engineering.md", "json": "https://newruntime.com/topics/context-engineering.json" }, "answer": [ "Context engineering is the work of turning scattered organizational knowledge into durable inputs an agent can inspect and act on.", "The strongest systems separate reusable rules, task state, source evidence, and operator corrections instead of burying everything in one prompt.", "The practical payoff is lower context waste and more reproducible agent behavior across long-running tasks." ] }, { "slug": "evals", "label": "Agent evals", "description": "A record of evals moving from model benchmarks toward behavior, workflow completion, skill reliability, and review burden.", "status": "featured", "last_updated": "2026-07-16", "counts": { "total": 13, "signals": 11, "patterns": 2, "posts": 0, "atlas": 0, "sources": 22 }, "routes": { "html": "https://newruntime.com/topics/evals/", "markdown": "https://newruntime.com/topics/evals.md", "json": "https://newruntime.com/topics/evals.json" }, "answer": [ "Agent evals need to measure behavior in a workflow, not only answer quality in a prompt.", "The relevant unit is completed work under constraints: tools, files, permissions, time, cost, and review outcome.", "A good eval makes regressions visible before the agent is trusted with larger work." ] }, { "slug": "agent-memory", "label": "Agent memory", "description": "Evidence on agent memory as a write loop: what should persist, who can correct it, and how it changes future work.", "status": "featured", "last_updated": "2026-07-21", "counts": { "total": 13, "signals": 9, "patterns": 1, "posts": 1, "atlas": 2, "sources": 19 }, "routes": { "html": "https://newruntime.com/topics/agent-memory/", "markdown": "https://newruntime.com/topics/agent-memory.md", "json": "https://newruntime.com/topics/agent-memory.json" }, "answer": [ "Agent memory is not only retrieval; it is a governed write loop that decides what survives a task.", "Useful memory needs correction-time updates, provenance, decay, and behavioral evaluation.", "The failure mode is treating memory as a passive store while the actual system behavior drifts." ] }, { "slug": "skills", "label": "Agent skills", "description": "Portable skill files, behavioral contracts, tool instructions, and evals that make agent capability reusable.", "status": "featured", "last_updated": "2026-07-04", "counts": { "total": 13, "signals": 13, "patterns": 0, "posts": 0, "atlas": 0, "sources": 22 }, "routes": { "html": "https://newruntime.com/topics/skills/", "markdown": "https://newruntime.com/topics/skills.md", "json": "https://newruntime.com/topics/skills.json" }, "answer": [ "Agent skills package repeatable behavior outside the current chat.", "The important boundary is whether the skill changes task outcomes, not whether the instruction prose looks good.", "Skills become infrastructure when they are versioned, evaluated, and shared across agents." ] }, { "slug": "model-routing", "label": "Model routing", "description": "How teams route work across model families, effort levels, cheap executors, judges, and local/runtime constraints.", "status": "featured", "last_updated": "2026-07-16", "counts": { "total": 13, "signals": 11, "patterns": 2, "posts": 0, "atlas": 0, "sources": 25 }, "routes": { "html": "https://newruntime.com/topics/model-routing/", "markdown": "https://newruntime.com/topics/model-routing.md", "json": "https://newruntime.com/topics/model-routing.json" }, "answer": [ "Model routing treats models as swappable execution resources rather than a single product choice.", "The pattern is to match task risk, latency, cost, and verification needs to the right model or model sequence.", "This makes the harness more durable than any one frontier model release." ] }, { "slug": "orchestration", "label": "Orchestration", "description": "How agent work is split into queues, roles, subagents, async loops, handoffs, and operator-visible state.", "status": "featured", "last_updated": "2026-07-15", "counts": { "total": 12, "signals": 12, "patterns": 0, "posts": 0, "atlas": 0, "sources": 17 }, "routes": { "html": "https://newruntime.com/topics/orchestration/", "markdown": "https://newruntime.com/topics/orchestration.md", "json": "https://newruntime.com/topics/orchestration.json" }, "answer": [ "Agent orchestration is the control layer around multiple autonomous or semi-autonomous work loops.", "It becomes necessary when one prompt cannot hold planning, execution, review, memory, and user coordination at once.", "The visible product shift is from a chat box to an operator surface for supervising work." ] }, { "slug": "agent-runtime", "label": "Agent runtime", "description": "The runtime layer around agents: sandboxes, tools, hosted execution, permissions, state, and deployment surfaces.", "status": "featured", "last_updated": "2026-07-22", "counts": { "total": 11, "signals": 9, "patterns": 1, "posts": 1, "atlas": 0, "sources": 23 }, "routes": { "html": "https://newruntime.com/topics/agent-runtime/", "markdown": "https://newruntime.com/topics/agent-runtime.md", "json": "https://newruntime.com/topics/agent-runtime.json" }, "answer": [ "The agent runtime is the system that lets models safely use tools, files, browsers, terminals, and external services.", "As agents take real actions, runtime concerns become product concerns: isolation, permissions, logging, replay, and rollback.", "This is where demos turn into repeatable infrastructure." ] }, { "slug": "verification", "label": "Verification", "description": "Signals and patterns about verification as the limiting factor for agent-produced work and AI-native operations.", "status": "featured", "last_updated": "2026-07-21", "counts": { "total": 10, "signals": 9, "patterns": 0, "posts": 1, "atlas": 0, "sources": 13 }, "routes": { "html": "https://newruntime.com/topics/verification/", "markdown": "https://newruntime.com/topics/verification.md", "json": "https://newruntime.com/topics/verification.json" }, "answer": [ "Verification is becoming the scarce resource in agentic work.", "More generation only helps when review, tests, traces, and provenance can keep up.", "The best systems make verification part of the product surface instead of an afterthought." ] }, { "slug": "agent-economics", "label": "Agent economics", "description": "How AI work is priced, measured, routed, and justified by completed task value rather than raw token volume.", "status": "featured", "last_updated": "2026-07-22", "counts": { "total": 8, "signals": 6, "patterns": 1, "posts": 1, "atlas": 0, "sources": 9 }, "routes": { "html": "https://newruntime.com/topics/agent-economics/", "markdown": "https://newruntime.com/topics/agent-economics.md", "json": "https://newruntime.com/topics/agent-economics.json" }, "answer": [ "Agent economics moves measurement from token use to useful completed work.", "Cost only makes sense beside success rate, review time, retry loops, latency, and human attention.", "The mature metric is cost per accepted outcome, not cost per generation." ] }, { "slug": "automation", "label": "Automation", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about automation.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 9, "signals": 7, "patterns": 1, "posts": 0, "atlas": 1, "sources": 15 }, "routes": { "html": "https://newruntime.com/topics/automation/", "markdown": "https://newruntime.com/topics/automation.md", "json": "https://newruntime.com/topics/automation.json" }, "answer": [ "Automation is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "codex", "label": "Codex", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about codex.", "status": "tracked", "last_updated": "2026-07-10", "counts": { "total": 8, "signals": 8, "patterns": 0, "posts": 0, "atlas": 0, "sources": 13 }, "routes": { "html": "https://newruntime.com/topics/codex/", "markdown": "https://newruntime.com/topics/codex.md", "json": "https://newruntime.com/topics/codex.json" }, "answer": [ "Codex is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "tool-use", "label": "Tool USE", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about tool use.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 8, "signals": 6, "patterns": 0, "posts": 1, "atlas": 1, "sources": 18 }, "routes": { "html": "https://newruntime.com/topics/tool-use/", "markdown": "https://newruntime.com/topics/tool-use.md", "json": "https://newruntime.com/topics/tool-use.json" }, "answer": [ "Tool USE is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "agent-ops", "label": "Agent OPS", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about agent ops.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 7, "signals": 2, "patterns": 0, "posts": 3, "atlas": 2, "sources": 12 }, "routes": { "html": "https://newruntime.com/topics/agent-ops/", "markdown": "https://newruntime.com/topics/agent-ops.md", "json": "https://newruntime.com/topics/agent-ops.json" }, "answer": [ "Agent OPS is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "agents", "label": "Agents", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about agents.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 7, "signals": 3, "patterns": 0, "posts": 2, "atlas": 2, "sources": 7 }, "routes": { "html": "https://newruntime.com/topics/agents/", "markdown": "https://newruntime.com/topics/agents.md", "json": "https://newruntime.com/topics/agents.json" }, "answer": [ "Agents is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "claude-code", "label": "Claude Code", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about claude code.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 7, "signals": 4, "patterns": 0, "posts": 3, "atlas": 0, "sources": 10 }, "routes": { "html": "https://newruntime.com/topics/claude-code/", "markdown": "https://newruntime.com/topics/claude-code.md", "json": "https://newruntime.com/topics/claude-code.json" }, "answer": [ "Claude Code is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "design-systems", "label": "Design Systems", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about design systems.", "status": "tracked", "last_updated": "2026-07-04", "counts": { "total": 7, "signals": 6, "patterns": 1, "posts": 0, "atlas": 0, "sources": 10 }, "routes": { "html": "https://newruntime.com/topics/design-systems/", "markdown": "https://newruntime.com/topics/design-systems.md", "json": "https://newruntime.com/topics/design-systems.json" }, "answer": [ "Design Systems is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "generative-ui", "label": "Generative UI", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about generative ui.", "status": "tracked", "last_updated": "2026-07-16", "counts": { "total": 7, "signals": 6, "patterns": 1, "posts": 0, "atlas": 0, "sources": 6 }, "routes": { "html": "https://newruntime.com/topics/generative-ui/", "markdown": "https://newruntime.com/topics/generative-ui.md", "json": "https://newruntime.com/topics/generative-ui.json" }, "answer": [ "Generative UI is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "mcp", "label": "MCP", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about mcp.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 7, "signals": 4, "patterns": 0, "posts": 1, "atlas": 2, "sources": 10 }, "routes": { "html": "https://newruntime.com/topics/mcp/", "markdown": "https://newruntime.com/topics/mcp.md", "json": "https://newruntime.com/topics/mcp.json" }, "answer": [ "MCP is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "multi-agent", "label": "Multi Agent", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about multi agent.", "status": "tracked", "last_updated": "2026-07-07", "counts": { "total": 7, "signals": 7, "patterns": 0, "posts": 0, "atlas": 0, "sources": 10 }, "routes": { "html": "https://newruntime.com/topics/multi-agent/", "markdown": "https://newruntime.com/topics/multi-agent.md", "json": "https://newruntime.com/topics/multi-agent.json" }, "answer": [ "Multi Agent is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "agent-loops", "label": "Agent Loops", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about agent loops.", "status": "tracked", "last_updated": "2026-07-14", "counts": { "total": 6, "signals": 5, "patterns": 1, "posts": 0, "atlas": 0, "sources": 11 }, "routes": { "html": "https://newruntime.com/topics/agent-loops/", "markdown": "https://newruntime.com/topics/agent-loops.md", "json": "https://newruntime.com/topics/agent-loops.json" }, "answer": [ "Agent Loops is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "agent-security", "label": "Agent Security", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about agent security.", "status": "tracked", "last_updated": "2026-07-22", "counts": { "total": 6, "signals": 3, "patterns": 0, "posts": 3, "atlas": 0, "sources": 8 }, "routes": { "html": "https://newruntime.com/topics/agent-security/", "markdown": "https://newruntime.com/topics/agent-security.md", "json": "https://newruntime.com/topics/agent-security.json" }, "answer": [ "Agent Security is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "enterprise-ai", "label": "Enterprise AI", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about enterprise ai.", "status": "tracked", "last_updated": "2026-07-16", "counts": { "total": 6, "signals": 6, "patterns": 0, "posts": 0, "atlas": 0, "sources": 8 }, "routes": { "html": "https://newruntime.com/topics/enterprise-ai/", "markdown": "https://newruntime.com/topics/enterprise-ai.md", "json": "https://newruntime.com/topics/enterprise-ai.json" }, "answer": [ "Enterprise AI is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "harness-engineering", "label": "Harness Engineering", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about harness engineering.", "status": "tracked", "last_updated": "2026-07-16", "counts": { "total": 6, "signals": 5, "patterns": 1, "posts": 0, "atlas": 0, "sources": 11 }, "routes": { "html": "https://newruntime.com/topics/harness-engineering/", "markdown": "https://newruntime.com/topics/harness-engineering.md", "json": "https://newruntime.com/topics/harness-engineering.json" }, "answer": [ "Harness Engineering is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "agent-skills", "label": "Agent Skills", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about agent skills.", "status": "tracked", "last_updated": "2026-07-16", "counts": { "total": 5, "signals": 4, "patterns": 1, "posts": 0, "atlas": 0, "sources": 10 }, "routes": { "html": "https://newruntime.com/topics/agent-skills/", "markdown": "https://newruntime.com/topics/agent-skills.md", "json": "https://newruntime.com/topics/agent-skills.json" }, "answer": [ "Agent Skills is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "api-design", "label": "API Design", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about api design.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 5, "signals": 2, "patterns": 1, "posts": 1, "atlas": 1, "sources": 6 }, "routes": { "html": "https://newruntime.com/topics/api-design/", "markdown": "https://newruntime.com/topics/api-design.md", "json": "https://newruntime.com/topics/api-design.json" }, "answer": [ "API Design is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "code-review", "label": "Code Review", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about code review.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 5, "signals": 4, "patterns": 0, "posts": 0, "atlas": 1, "sources": 6 }, "routes": { "html": "https://newruntime.com/topics/code-review/", "markdown": "https://newruntime.com/topics/code-review.md", "json": "https://newruntime.com/topics/code-review.json" }, "answer": [ "Code Review is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "engineering-management", "label": "Engineering Management", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about engineering management.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 5, "signals": 2, "patterns": 2, "posts": 1, "atlas": 0, "sources": 13 }, "routes": { "html": "https://newruntime.com/topics/engineering-management/", "markdown": "https://newruntime.com/topics/engineering-management.md", "json": "https://newruntime.com/topics/engineering-management.json" }, "answer": [ "Engineering Management is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "enterprise-agents", "label": "Enterprise Agents", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about enterprise agents.", "status": "tracked", "last_updated": "2026-07-21", "counts": { "total": 5, "signals": 4, "patterns": 0, "posts": 1, "atlas": 0, "sources": 6 }, "routes": { "html": "https://newruntime.com/topics/enterprise-agents/", "markdown": "https://newruntime.com/topics/enterprise-agents.md", "json": "https://newruntime.com/topics/enterprise-agents.json" }, "answer": [ "Enterprise Agents is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "loop-engineering", "label": "Loop Engineering", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about loop engineering.", "status": "tracked", "last_updated": "2026-07-16", "counts": { "total": 5, "signals": 4, "patterns": 0, "posts": 1, "atlas": 0, "sources": 7 }, "routes": { "html": "https://newruntime.com/topics/loop-engineering/", "markdown": "https://newruntime.com/topics/loop-engineering.md", "json": "https://newruntime.com/topics/loop-engineering.json" }, "answer": [ "Loop Engineering is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "personalization", "label": "Personalization", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about personalization.", "status": "tracked", "last_updated": "2026-07-15", "counts": { "total": 5, "signals": 5, "patterns": 0, "posts": 0, "atlas": 0, "sources": 5 }, "routes": { "html": "https://newruntime.com/topics/personalization/", "markdown": "https://newruntime.com/topics/personalization.md", "json": "https://newruntime.com/topics/personalization.json" }, "answer": [ "Personalization is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "retrieval", "label": "Retrieval", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about retrieval.", "status": "tracked", "last_updated": "2026-07-15", "counts": { "total": 5, "signals": 5, "patterns": 0, "posts": 0, "atlas": 0, "sources": 6 }, "routes": { "html": "https://newruntime.com/topics/retrieval/", "markdown": "https://newruntime.com/topics/retrieval.md", "json": "https://newruntime.com/topics/retrieval.json" }, "answer": [ "Retrieval is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] }, { "slug": "sandboxes", "label": "Sandboxes", "description": "A New Runtime topic hub collecting signals, patterns, field notes, and public sources about sandboxes.", "status": "tracked", "last_updated": "2026-07-14", "counts": { "total": 5, "signals": 5, "patterns": 0, "posts": 0, "atlas": 0, "sources": 8 }, "routes": { "html": "https://newruntime.com/topics/sandboxes/", "markdown": "https://newruntime.com/topics/sandboxes.md", "json": "https://newruntime.com/topics/sandboxes.json" }, "answer": [ "Sandboxes is tracked here as an evidence-linked topic, not as a static glossary entry.", "The page connects raw observations to pattern hypotheses, longer analysis, and public sources.", "Use it as the canonical landing page before drilling into individual records." ] } ], "featured": [ { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "coding-agents", "title": "Coding agents - New Runtime", "description": "A living map of coding agents as they move from autocomplete into supervised software work, review loops, environments, and operating cost.", "answer": [ "Coding agents are becoming operational workers inside software teams, not just better code completion.", "The durable bottlenecks are review bandwidth, environment setup, context quality, permissions, and cost per completed task.", "The useful question is no longer whether an agent can write code, but whether the surrounding harness can verify and land work repeatedly." ], "search_intents": [ "coding agent workflow", "AI coding agent verification", "software engineering agents" ], "status": "featured", "last_updated": "2026-07-22", "counts": { "total": 49, "signals": 39, "patterns": 1, "posts": 6, "atlas": 3, "sources": 81 }, "routes": { "html": "https://newruntime.com/topics/coding-agents/", "markdown": "https://newruntime.com/topics/coding-agents.md", "json": "https://newruntime.com/topics/coding-agents.json" }, "top_sources": [ "https://addyo.substack.com/p/own-the-outer-loop", "https://addyosmani.com/blog/new-sdlc-vibe-coding", "https://agent-cookbook.com/tutorial/ai-20260512163240-54x4", "https://agentos-sdk.dev/", "https://anthropic.com/news/claude-sonnet-5", "https://anthropic.com/research/AI-assistance-coding-skills", "https://anthropic.com/research/claude-code-expertise", "https://arxiv.org/abs/2207.06991", "https://arxiv.org/abs/2412.09871", "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/", "https://claude.com/blog/agent-view-in-claude-code", "https://claude.com/blog/artifacts-in-claude-code" ], "patterns": [ { "kind": "Pattern", "slug": "verification-bandwidth-is-the-scarce-resource", "title": "Verification bandwidth is the scarce engineering resource", "description": "The primary bottleneck in agentic software delivery is moving from code production to the human and machine capacity required to verify it.", "date": "2026-07-16", "topics": [ "coding-agents", "evals", "engineering-management" ], "source_count": 5, "metric": "high", "url": "https://newruntime.com/patterns/verification-bandwidth-is-the-scarce-resource/" } ], "field_notes": [ { "kind": "Field Note", "slug": "devin-outposts-agent-execution-plane", "title": "Devin Outposts Splits the Agent Brain from the Execution Plane", "description": "Devin Outposts moves command execution, repository access, and sandbox lifecycle into customer-controlled infrastructure while the agent loop remains in Cognition's cloud.", "date": "2026-07-22", "topics": [ "coding-agents", "agent-runtime", "infrastructure" ], "source_count": 4, "url": "https://newruntime.com/posts/devin-outposts-agent-execution-plane/" }, { "kind": "Field Note", "slug": "gemini-3-6-flash-agent-efficiency", "title": "Gemini 3.6 Flash Moves the Agent Race Toward Cost per Task", "description": "Google's Gemini 3.6 Flash release frames the model race around token efficiency, built-in computer use, and specialized cyber agents rather than raw chat intelligence alone.", "date": "2026-07-22", "topics": [ "gemini", "agent-economics", "coding-agents" ], "source_count": 1, "url": "https://newruntime.com/posts/gemini-3-6-flash-agent-efficiency/" }, { "kind": "Field Note", "slug": "ai-coding-workflow-verifiable-work", "title": "AI Coding Workflow: From Idea to Verifiable Work", "description": "AI coding works better when checkable artifacts stand between the idea and the code: specs, tickets, TDD, fresh-context review, and manual QA.", "date": "2026-07-21", "topics": [ "coding-agents", "workflow", "verification" ], "source_count": 2, "url": "https://newruntime.com/posts/ai-coding-workflow-verifiable-work/" }, { "kind": "Field Note", "slug": "coding-agent-cost-environment-config", "title": "Coding Agent Cost Is Cut in Environment Config, Not Prompts", "description": "Claude Code and Codex costs are reduced by environment design, not by asking the agent to read less: output filtering, repo maps, model routing, and stable prompt caching.", "date": "2026-07-21", "topics": [ "coding-agents", "cost-control", "prompt-caching" ], "source_count": 3, "url": "https://newruntime.com/posts/coding-agent-cost-environment-config/" }, { "kind": "Field Note", "slug": "pillar-sandbox-escapes", "title": "Coding Agent Sandboxes Break in Places Teams Do Not Expect", "description": "Pillar shows that agent sandboxes must be assessed not only around the agent process, but around files, configs, allowlisted commands, and local daemons the host later trusts.", "date": "2026-07-21", "topics": [ "agent-security", "coding-agents", "sandbox" ], "source_count": 1, "url": "https://newruntime.com/posts/pillar-sandbox-escapes/" }, { "kind": "Field Note", "slug": "kimi-k3-frontend-model-race", "title": "Kimi K3: Frontend Is Becoming the Model Race Arena", "description": "Kimi K3 moves open model competition into visual software engineering: frontend benchmarks require not only code, but layout, screenshots, accessibility, and human preference.", "date": "2026-07-21", "topics": [ "open-models", "frontend", "coding-agents" ], "source_count": 2, "url": "https://newruntime.com/posts/kimi-k3-frontend-model-race/" } ], "raw_signals": [ { "kind": "Raw Signal", "slug": "opencode-model-swappable-agent-harness", "title": "OpenCode makes the model a swappable dependency", "description": "OpenCode keeps rules, skills, permissions, sessions, and tools stable while the underlying model changes.", "date": "2026-07-16", "topics": [ "coding-agents", "harness-engineering", "model-routing", "developer-tools" ], "source_count": 4, "metric": "structural", "url": "https://newruntime.com/signals/opencode-model-swappable-agent-harness/" }, { "kind": "Raw Signal", "slug": "verification-capacity-coding-agent-bottleneck", "title": "Verification capacity becomes the coding-agent bottleneck", "description": "Coding agents can parallelize implementation faster than engineering teams can expand review judgment.", "date": "2026-07-16", "topics": [ "coding-agents", "evals", "agent-ops", "engineering-management" ], "source_count": 4, "metric": "structural", "url": "https://newruntime.com/signals/verification-capacity-coding-agent-bottleneck/" }, { "kind": "Raw Signal", "slug": "fable-cost-per-completed-task", "title": "A pricier model can be cheaper per completed task", "description": "Cognition reports that Fable 5 completed coding work with fewer steps and output tokens than its previous lead model.", "date": "2026-07-15", "topics": [ "agent-economics", "coding-agents", "model-routing" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/fable-cost-per-completed-task/" }, { "kind": "Raw Signal", "slug": "airbnb-agentic-coding-loop", "title": "Airbnb frames agentic coding as a loop inside development", "description": "The workflow separates planning, delegated implementation, verification, and integration instead of treating the coding agent as the whole process.", "date": "2026-07-14", "topics": [ "coding-agents", "developer-workflow", "agent-loops" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/airbnb-agentic-coding-loop/" }, { "kind": "Raw Signal", "slug": "claude-code-model-and-effort-are-separate-controls", "title": "Claude Code separates model choice from effort", "description": "Anthropic exposes model selection and effort level as different controls for capability, token use, latency, and persistence.", "date": "2026-07-13", "topics": [ "coding-agents", "model-routing", "agent-economics" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/claude-code-model-and-effort-are-separate-controls/" }, { "kind": "Raw Signal", "slug": "cheap-code-moves-bottleneck-to-review", "title": "Cheap code moves the engineering bottleneck to review", "description": "An agent-heavy development model treats implementation as abundant while specifications, validation, security, and integration remain scarce.", "date": "2026-07-12", "topics": [ "engineering-management", "coding-agents", "verification" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/cheap-code-moves-bottleneck-to-review/" }, { "kind": "Raw Signal", "slug": "databricks-benchmarks-real-coding-agent-economics", "title": "Databricks benchmarks coding agents on its own codebase", "description": "Databricks evaluates agents on fresh internal pull-request tasks and measures success alongside runtime, tokens, and cost.", "date": "2026-07-12", "topics": [ "coding-agents", "evals", "agent-economics" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/databricks-benchmarks-real-coding-agent-economics/" }, { "kind": "Raw Signal", "slug": "large-task-planning-becomes-ticket-graph", "title": "Large-task planning moves from chat into a ticket graph", "description": "Matt Pocock's workflow turns discovery, specification, tickets, implementation, and review into durable linked artifacts.", "date": "2026-07-12", "topics": [ "coding-agents", "planning", "task-graphs", "agent-skills" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/large-task-planning-becomes-ticket-graph/" }, { "kind": "Raw Signal", "slug": "an-ai-coding-workflow-starts-with-data-and-interfaces", "title": "An AI Coding Workflow Starts with Data and Interfaces", "description": "ThePrimeagen's workflow separates evidence gathering, interface design, and implementation so the coding agent receives a constrained problem instead of an open-ended prompt.", "date": "2026-07-11", "topics": [ "coding-agents", "specification", "workflow" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/an-ai-coding-workflow-starts-with-data-and-interfaces/" }, { "kind": "Raw Signal", "slug": "trycase-disposable-linux-agent-verification", "title": "TryCase gives coding agents disposable Linux verification", "description": "TryCase creates short-lived Linux environments where agents can run generated code and tests away from the developer's machine.", "date": "2026-07-10", "topics": [ "sandboxes", "coding-agents", "verification" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/trycase-disposable-linux-agent-verification/" }, { "kind": "Raw Signal", "slug": "parallel-agents-make-large-migrations-tractable", "title": "Parallel coding agents make large migrations tractable", "description": "A large Bun pull request shows many independent agents working against explicit file boundaries and a common test contract.", "date": "2026-07-08", "topics": [ "coding-agents", "parallel-agents", "software-migrations" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/parallel-agents-make-large-migrations-tractable/" }, { "kind": "Raw Signal", "slug": "pxpipe-trades-text-tokens-for-rendered-visual-context", "title": "pxpipe Trades Text Tokens for Rendered Visual Context", "description": "The proxy compresses large tool results and old context into images, reducing input cost while exposing precision limits for byte-exact data.", "date": "2026-07-08", "topics": [ "context-compression", "token-cost", "coding-agents" ], "source_count": 4, "metric": "notable", "url": "https://newruntime.com/signals/pxpipe-trades-text-tokens-for-rendered-visual-context/" }, { "kind": "Raw Signal", "slug": "claude-code-recovers-after-failed-command", "title": "Claude Code can recover after a failed command", "description": "A runtime update lets the coding agent inspect a command failure and continue its loop without waiting for a generic human prompt.", "date": "2026-07-07", "topics": [ "coding-agents", "error-recovery", "agent-loops" ], "source_count": 1, "metric": "incremental", "url": "https://newruntime.com/signals/claude-code-recovers-after-failed-command/" }, { "kind": "Raw Signal", "slug": "openwiki-codebase-docs-as-agent-memory", "title": "OpenWiki treats repository documentation as agent memory", "description": "OpenWiki generates and refreshes a structured codebase explanation that other agents and humans can inspect before acting.", "date": "2026-07-07", "topics": [ "codebase-memory", "documentation", "coding-agents" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/openwiki-codebase-docs-as-agent-memory/" }, { "kind": "Raw Signal", "slug": "herdr-adds-agent-awareness-to-terminal-multiplexing", "title": "Herdr Adds Agent Awareness to Terminal Multiplexing", "description": "Herdr organizes parallel coding-agent sessions around projects, process state, and operator navigation rather than presenting them as undifferentiated terminal panes.", "date": "2026-07-04", "topics": [ "coding-agents", "orchestration", "terminal" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/herdr-adds-agent-awareness-to-terminal-multiplexing/" }, { "kind": "Raw Signal", "slug": "twenty-four-gigabytes-of-vram-becomes-a-practical-local-coding-tier", "title": "Twenty-Four Gigabytes of VRAM Becomes a Practical Local Coding Tier", "description": "Local coding models are becoming useful on high-end consumer hardware, creating a viable private lane even when frontier cloud agents remain stronger overall.", "date": "2026-07-04", "topics": [ "local-models", "coding-agents", "hardware" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/twenty-four-gigabytes-of-vram-becomes-a-practical-local-coding-tier/" }, { "kind": "Raw Signal", "slug": "sonnet-moves-agent-capability-downmarket", "title": "Sonnet moves agent capability down the price curve", "description": "Anthropic positions Claude Sonnet 5 for planning, terminal work, browser use, and multi-step agent tasks at a lower tier.", "date": "2026-07-01", "topics": [ "models", "agent-economics", "coding-agents" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/sonnet-moves-agent-capability-downmarket/" }, { "kind": "Raw Signal", "slug": "warp-reframes-product-engineers-as-factory-engineers", "title": "Warp Reframes Product Engineers as Factory Engineers", "description": "An AI-native engineering organization invests more effort in task decomposition, agent infrastructure, review systems, and throughput management than in direct implementation.", "date": "2026-06-30", "topics": [ "engineering-org", "coding-agents", "orchestration" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/warp-reframes-product-engineers-as-factory-engineers/" }, { "kind": "Raw Signal", "slug": "agentos-lightweight-runtime-for-coding-agents", "title": "agentOS packages a lightweight runtime for coding agents", "description": "Rivet's agentOS combines isolated execution, durable workflows, resource limits, and host-side secrets in a WebAssembly and Rust runtime.", "date": "2026-06-29", "topics": [ "agent-runtime", "sandboxes", "coding-agents" ], "source_count": 3, "metric": "notable", "url": "https://newruntime.com/signals/agentos-lightweight-runtime-for-coding-agents/" }, { "kind": "Raw Signal", "slug": "loop-library-packages-fifteen-repeatable-coding-agent-cycles", "title": "Loop Library Packages Fifteen Repeatable Coding-Agent Cycles", "description": "A public repository turns common maintenance and delivery tasks into explicit loops with commands, checks, and completion conditions.", "date": "2026-06-28", "topics": [ "loop-engineering", "coding-agents", "automation" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/loop-library-packages-fifteen-repeatable-coding-agent-cycles/" }, { "kind": "Raw Signal", "slug": "shopify-treats-code-as-output-and-guardrails-as-engineering", "title": "Shopify Treats Code as Output and Guardrails as Engineering", "description": "The engineering role shifts toward defining constraints, interfaces, tests, and product intent while agents perform a larger share of implementation.", "date": "2026-06-28", "topics": [ "engineering-org", "guardrails", "coding-agents" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/shopify-treats-code-as-output-and-guardrails-as-engineering/" }, { "kind": "Raw Signal", "slug": "cursor-is-expanding-from-ide-to-agent-factory", "title": "Cursor Is Expanding from IDE to Agent Factory", "description": "Cursor's workflow increasingly centers on parallel autonomous tasks, remote execution, review, and orchestration rather than one developer editing one file at a time.", "date": "2026-06-27", "topics": [ "cursor", "coding-agents", "orchestration" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/cursor-is-expanding-from-ide-to-agent-factory/" }, { "kind": "Raw Signal", "slug": "glm-5-2-changes-the-economics-of-claude-code-workflows", "title": "GLM-5.2 Changes the Economics of Claude Code Workflows", "description": "A capable open model routed through compatible infrastructure can replace premium inference for parts of a coding-agent workload.", "date": "2026-06-27", "topics": [ "open-models", "model-routing", "coding-agents" ], "source_count": 4, "metric": "notable", "url": "https://newruntime.com/signals/glm-5-2-changes-the-economics-of-claude-code-workflows/" }, { "kind": "Raw Signal", "slug": "parallel-agent-workflow-needs-operator-stack", "title": "Parallel coding agents need an operator stack", "description": "Kun Chen's workflow combines durable project context, task plans, worktrees, independent review, tests, and cleanup around multiple agents.", "date": "2026-06-27", "topics": [ "coding-agents", "parallel-agents", "developer-workflow" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/parallel-agent-workflow-needs-operator-stack/" } ], "atlas_records": [ { "kind": "Project", "slug": "coding-agent-cost-controls", "title": "Coding Agent Cost Controls", "description": "A forkable setup for reducing coding-agent cost by shaping environment output, repo maps, model routing, cache stability, and task-level budgets.", "date": "2026-07-21", "topics": [ "coding-agents", "cost-control" ], "source_count": 2, "metric": "B2", "url": "https://newruntime.com/projects/coding-agent-cost-controls/" }, { "kind": "Shift", "slug": "goal-scoped-agent-loops", "title": "Manual continue prompting -> goal-scoped agent loops", "description": "Long-running agent work is moving from manual continue prompts toward explicit goals, limits, subagents, schedules, and verifiable completion.", "date": "2026-07-21", "topics": [ "coding-agents", "agent-ops", "automation" ], "source_count": 3, "metric": "emerging", "url": "https://newruntime.com/shifts/goal-scoped-agent-loops/" }, { "kind": "Project", "slug": "verifiable-ai-coding-workflow", "title": "Verifiable AI Coding Workflow", "description": "A small agent-assisted coding workflow where the useful artifact is not generated code, but a checked path from idea to tests, diff, review, and handoff.", "date": "2026-07-21", "topics": [ "coding-agents", "developer-workflow" ], "source_count": 1, "metric": "B2", "url": "https://newruntime.com/projects/verifiable-ai-coding-workflow/" } ] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "context-engineering", "title": "Context engineering - New Runtime", "description": "How teams package memory, rules, traces, tools, and task state so agents can work without restarting from raw chat context.", "answer": [ "Context engineering is the work of turning scattered organizational knowledge into durable inputs an agent can inspect and act on.", "The strongest systems separate reusable rules, task state, source evidence, and operator corrections instead of burying everything in one prompt.", "The practical payoff is lower context waste and more reproducible agent behavior across long-running tasks." ], "search_intents": [ "context engineering", "agent context architecture", "LLM context systems" ], "status": "featured", "last_updated": "2026-07-21", "counts": { "total": 16, "signals": 9, "patterns": 1, "posts": 3, "atlas": 3, "sources": 27 }, "routes": { "html": "https://newruntime.com/topics/context-engineering/", "markdown": "https://newruntime.com/topics/context-engineering.md", "json": "https://newruntime.com/topics/context-engineering.json" }, "top_sources": [ "https://aclanthology.org/2024.acl-long.747", "https://aihero.dev/how-to-kill-the-bloat-in-claude-codes-system-prompt", "https://arxiv.org/abs/2410.10813", "https://arxiv.org/abs/2602.12670", "https://arxiv.org/abs/2603.29919", "https://arxiv.org/abs/2605.24050", "https://claude.com/blog/steering-claude-code-skills-hooks-rules-subagents-and-more", "https://code.claude.com/docs/en/sub-agents", "https://dropbox.tech/security/dropbox-mcp-dash-design-code-security", "https://engineering.atspotify.com/2026/6/encoding-your-domain-expert-the-context-layer-behind-spotifys-data-assistant", "https://engineering.grab.com/scaling-out-distroless-adoption-with-ai", "https://github.com/AI-Builder-Club/skills" ], "patterns": [ { "kind": "Pattern", "slug": "skills-become-portable-capability-layer", "title": "Skills become a portable capability layer", "description": "Agent skills are emerging as a portable capability layer, but their value depends on progressive disclosure, provenance, security review, and behavioral evaluation.", "date": "2026-07-16", "topics": [ "agent-skills", "evals", "context-engineering" ], "source_count": 4, "metric": "medium", "url": "https://newruntime.com/patterns/skills-become-portable-capability-layer/" } ], "field_notes": [ { "kind": "Field Note", "slug": "ai-native-engineering-team", "title": "AI-Native Engineering Teams Are Smaller and Stricter", "description": "AI-native engineering changes team size, roles, onboarding, metrics, and review: code is no longer written only by humans, but context and quality ownership get stricter.", "date": "2026-07-21", "topics": [ "ai-native-teams", "engineering-management", "context-engineering" ], "source_count": 1, "url": "https://newruntime.com/posts/ai-native-engineering-team/" }, { "kind": "Field Note", "slug": "claude-code-subagents-memory", "title": "Claude Code Subagents Split Memory Instead of Inflating One Session", "description": "Claude Code subagents show a practical form of agent memory hygiene: move noisy work into a separate context and return only the compressed result to the main session.", "date": "2026-07-21", "topics": [ "claude-code", "subagents", "context-engineering" ], "source_count": 1, "url": "https://newruntime.com/posts/claude-code-subagents-memory/" }, { "kind": "Field Note", "slug": "unabyss-shared-context-layer", "title": "Unabyss Builds a Shared Context Layer Across Claude, Codex, and Cursor", "description": "Unabyss points to where agent tooling is going: context stops being one client's internal memory and becomes a separate product layer with governance and portability.", "date": "2026-07-21", "topics": [ "context-engineering", "mcp", "agent-tools" ], "source_count": 2, "url": "https://newruntime.com/posts/unabyss-shared-context-layer/" } ], "raw_signals": [ { "kind": "Raw Signal", "slug": "agent-skills-need-behavioral-evals", "title": "Agent skills need behavioral evals, not prose review", "description": "Benchmarks show that expert-authored skills can help while self-generated skills can underperform a no-skill baseline.", "date": "2026-07-16", "topics": [ "agent-skills", "evals", "context-engineering" ], "source_count": 5, "metric": "structural", "url": "https://newruntime.com/signals/agent-skills-need-behavioral-evals/" }, { "kind": "Raw Signal", "slug": "retrieval-does-not-create-company-memory", "title": "Retrieval and long context do not create company memory", "description": "Finding a relevant fragment does not establish whether it is current, authorized, trustworthy, or applicable to the present task.", "date": "2026-07-15", "topics": [ "company-memory", "retrieval", "context-engineering" ], "source_count": 2, "metric": "structural", "url": "https://newruntime.com/signals/retrieval-does-not-create-company-memory/" }, { "kind": "Raw Signal", "slug": "claude-code-startup-context-can-be-reduced-to-13k-tokens", "title": "Claude Code Startup Context Can Be Reduced to 13K Tokens", "description": "A measured context audit shows that tool descriptions and default instructions consume a material share of the initial window before repository work begins.", "date": "2026-07-09", "topics": [ "claude-code", "context-engineering", "token-economics" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/claude-code-startup-context-can-be-reduced-to-13k-tokens/" }, { "kind": "Raw Signal", "slug": "codebase-memory-combines-search-with-a-repository-map", "title": "Codebase Memory Combines Search with a Repository Map", "description": "A code retrieval tool adds structural repository context to text search so an agent can locate relevant code without loading the whole project into context.", "date": "2026-07-03", "topics": [ "code-search", "agent-memory", "context-engineering" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/codebase-memory-combines-search-with-a-repository-map/" }, { "kind": "Raw Signal", "slug": "pydantic-ai-capabilities-progressive-disclosure", "title": "Pydantic AI loads capabilities only when needed", "description": "A capability can bundle instructions, tools, model settings, and hooks while exposing only a compact description until activation.", "date": "2026-06-28", "topics": [ "agent-skills", "context-engineering", "tool-use" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/pydantic-ai-capabilities-progressive-disclosure/" }, { "kind": "Raw Signal", "slug": "claude-md-shrinks-as-skills-hooks-and-subagents-mature", "title": "CLAUDE.md Shrinks as Skills, Hooks, and Subagents Mature", "description": "Agent behavior moves out of one giant context file into scoped mechanisms that load only for the relevant task and lifecycle stage.", "date": "2026-06-23", "topics": [ "claude-code", "skills", "context-engineering" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/claude-md-shrinks-as-skills-hooks-and-subagents-mature/" }, { "kind": "Raw Signal", "slug": "production-ai-engineering-is-the-system-around-the-model", "title": "Production AI Engineering Is the System Around the Model", "description": "Examples from large engineering teams point to testing, context layers, security, and workflow integration as the durable production work.", "date": "2026-06-16", "topics": [ "production-ai", "agent-evals", "context-engineering" ], "source_count": 5, "metric": "notable", "url": "https://newruntime.com/signals/production-ai-engineering-is-the-system-around-the-model/" }, { "kind": "Raw Signal", "slug": "codex-subagents-reduce-context-pollution", "title": "Codex Subagents Reduce Context Pollution", "description": "Delegating bounded work to subagents preserves the parent agent's attention and creates clearer evidence boundaries than loading every exploration step into one conversation.", "date": "2026-05-24", "topics": [ "codex", "subagents", "context-engineering" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/codex-subagents-reduce-context-pollution/" }, { "kind": "Raw Signal", "slug": "skill-md-acts-as-a-behavior-loader-not-a-better-prompt", "title": "SKILL.md Acts as a Behavior Loader, Not a Better Prompt", "description": "Skills package procedures, tools, and progressive context so an agent can load behavior only when the task requires it.", "date": "2026-05-07", "topics": [ "skills", "context-engineering", "agent-harness" ], "source_count": 3, "metric": "notable", "url": "https://newruntime.com/signals/skill-md-acts-as-a-behavior-loader-not-a-better-prompt/" } ], "atlas_records": [ { "kind": "Shift", "slug": "shared-context-infrastructure", "title": "Client-local memory -> shared context infrastructure", "description": "Context is shifting from private memory inside one AI client into a governed layer between agents, IDEs, documents, and tools.", "date": "2026-07-21", "topics": [ "context-engineering", "agent-memory", "mcp" ], "source_count": 2, "metric": "emerging", "url": "https://newruntime.com/shifts/shared-context-infrastructure/" }, { "kind": "Graveyard", "slug": "manual-context-handoff", "title": "Manual context handoff between AI clients", "description": "The ritual of retelling one AI client what another already understood becomes optional when context moves into a shared layer.", "date": "2026-07-21", "topics": [ "context-engineering", "agent-memory" ], "source_count": 1, "metric": "challenged", "url": "https://newruntime.com/graveyard/manual-context-handoff/" }, { "kind": "Delta", "slug": "shared-context-portability", "title": "Shared context lets AI clients stop starting from zero", "description": "A separate context layer lets multiple AI clients read shared project context without a human manually copying memory.", "date": "2026-07-21", "topics": [ "context-engineering", "mcp" ], "source_count": 2, "metric": "B1", "url": "https://newruntime.com/deltas/shared-context-portability/" } ] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "evals", "title": "Agent evals - New Runtime", "description": "A record of evals moving from model benchmarks toward behavior, workflow completion, skill reliability, and review burden.", "answer": [ "Agent evals need to measure behavior in a workflow, not only answer quality in a prompt.", "The relevant unit is completed work under constraints: tools, files, permissions, time, cost, and review outcome.", "A good eval makes regressions visible before the agent is trusted with larger work." ], "search_intents": [ "agent evals", "AI agent evaluation", "LLM workflow benchmarks" ], "status": "featured", "last_updated": "2026-07-16", "counts": { "total": 13, "signals": 11, "patterns": 2, "posts": 0, "atlas": 0, "sources": 22 }, "routes": { "html": "https://newruntime.com/topics/evals/", "markdown": "https://newruntime.com/topics/evals.md", "json": "https://newruntime.com/topics/evals.json" }, "top_sources": [ "https://addyo.substack.com/p/own-the-outer-loop", "https://addyosmani.com/blog/new-sdlc-vibe-coding", "https://anthropic.com/research/AI-assistance-coding-skills", "https://arxiv.org/abs/2602.12670", "https://arxiv.org/abs/2603.29919", "https://arxiv.org/abs/2605.24050", "https://databricks.com/blog/benchmarking-coding-agents-databricks-multi-million-line-codebase", "https://dbreunig.com/2026/06/22/the-problem-is-prompt-debt.html", "https://developers.openai.com/api/docs/guides/latest-model", "https://github.com/benchflow-ai/skillsbench", "https://github.com/OpenDataBox/MemoryData", "https://github.com/shadcn/improve" ], "patterns": [ { "kind": "Pattern", "slug": "skills-become-portable-capability-layer", "title": "Skills become a portable capability layer", "description": "Agent skills are emerging as a portable capability layer, but their value depends on progressive disclosure, provenance, security review, and behavioral evaluation.", "date": "2026-07-16", "topics": [ "agent-skills", "evals", "context-engineering" ], "source_count": 4, "metric": "medium", "url": "https://newruntime.com/patterns/skills-become-portable-capability-layer/" }, { "kind": "Pattern", "slug": "verification-bandwidth-is-the-scarce-resource", "title": "Verification bandwidth is the scarce engineering resource", "description": "The primary bottleneck in agentic software delivery is moving from code production to the human and machine capacity required to verify it.", "date": "2026-07-16", "topics": [ "coding-agents", "evals", "engineering-management" ], "source_count": 5, "metric": "high", "url": "https://newruntime.com/patterns/verification-bandwidth-is-the-scarce-resource/" } ], "field_notes": [], "raw_signals": [ { "kind": "Raw Signal", "slug": "agent-skills-need-behavioral-evals", "title": "Agent skills need behavioral evals, not prose review", "description": "Benchmarks show that expert-authored skills can help while self-generated skills can underperform a no-skill baseline.", "date": "2026-07-16", "topics": [ "agent-skills", "evals", "context-engineering" ], "source_count": 5, "metric": "structural", "url": "https://newruntime.com/signals/agent-skills-need-behavioral-evals/" }, { "kind": "Raw Signal", "slug": "verification-capacity-coding-agent-bottleneck", "title": "Verification capacity becomes the coding-agent bottleneck", "description": "Coding agents can parallelize implementation faster than engineering teams can expand review judgment.", "date": "2026-07-16", "topics": [ "coding-agents", "evals", "agent-ops", "engineering-management" ], "source_count": 4, "metric": "structural", "url": "https://newruntime.com/signals/verification-capacity-coding-agent-bottleneck/" }, { "kind": "Raw Signal", "slug": "ai-feedback-needs-behavioral-traces", "title": "AI product feedback needs behavioral traces", "description": "For nondeterministic products, edits, retries, overrides, abandonment, and sampled outputs reveal more than isolated ratings.", "date": "2026-07-14", "topics": [ "product-feedback", "evals", "human-traces" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/ai-feedback-needs-behavioral-traces/" }, { "kind": "Raw Signal", "slug": "databricks-benchmarks-real-coding-agent-economics", "title": "Databricks benchmarks coding agents on its own codebase", "description": "Databricks evaluates agents on fresh internal pull-request tasks and measures success alongside runtime, tokens, and cost.", "date": "2026-07-12", "topics": [ "coding-agents", "evals", "agent-economics" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/databricks-benchmarks-real-coding-agent-economics/" }, { "kind": "Raw Signal", "slug": "gpt-5-6-shorter-prompts", "title": "Newer agent models can reward shorter prompts", "description": "OpenAI's migration guidance emphasizes concise instructions, explicit contracts, and less compensating prose for newer models.", "date": "2026-07-10", "topics": [ "prompt-engineering", "model-migration", "evals" ], "source_count": 1, "metric": "incremental", "url": "https://newruntime.com/signals/gpt-5-6-shorter-prompts/" }, { "kind": "Raw Signal", "slug": "continuous-evals-for-multi-agent-systems", "title": "Multi-agent systems need continuous eval pipelines", "description": "A Google workflow evaluates not only the final answer but also routing, delegation, tool calls, and the trajectory between agents.", "date": "2026-07-07", "topics": [ "multi-agent", "evals", "observability" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/continuous-evals-for-multi-agent-systems/" }, { "kind": "Raw Signal", "slug": "agent-memory-evaluated-as-system-behavior", "title": "Agent memory is being evaluated as system behavior", "description": "MemoryData evaluates what an agent stores, retrieves, updates, and forgets across a sequence instead of grading one final answer.", "date": "2026-07-06", "topics": [ "agent-memory", "evals", "knowledge-systems" ], "source_count": 2, "metric": "structural", "url": "https://newruntime.com/signals/agent-memory-evaluated-as-system-behavior/" }, { "kind": "Raw Signal", "slug": "harvey-trains-legal-agent-with-applied-compute", "title": "Harvey trained a legal agent with an applied compute loop", "description": "Harvey describes domain experts, task environments, evaluation, and iterative training as one system for legal agent performance.", "date": "2026-07-03", "topics": [ "vertical-agents", "evals", "legal-ai" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/harvey-trains-legal-agent-with-applied-compute/" }, { "kind": "Raw Signal", "slug": "verifiability-is-an-ai-product-feature", "title": "Verifiability is an AI product feature", "description": "Hamel Husain's eval-smell framework treats missing traces, weak rubrics, and unverifiable outputs as product defects.", "date": "2026-07-02", "topics": [ "evals", "observability", "ai-product" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/verifiability-is-an-ai-product-feature/" }, { "kind": "Raw Signal", "slug": "prompt-debt-becomes-legacy-code", "title": "Prompt debt turns natural language into legacy code", "description": "System prompts accumulate patches for edge cases until model changes and local fixes make the behavior fragile and opaque.", "date": "2026-06-29", "topics": [ "prompt-engineering", "evals", "technical-debt" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/prompt-debt-becomes-legacy-code/" }, { "kind": "Raw Signal", "slug": "ai-compresses-coding-not-whole-sdlc", "title": "AI compresses coding, not the whole software lifecycle", "description": "Implementation time can collapse while requirements, architecture, validation, review, and maintenance stay constrained by human judgment.", "date": "2026-06-23", "topics": [ "software-lifecycle", "coding-agents", "evals" ], "source_count": 2, "metric": "structural", "url": "https://newruntime.com/signals/ai-compresses-coding-not-whole-sdlc/" } ], "atlas_records": [] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "agent-memory", "title": "Agent memory - New Runtime", "description": "Evidence on agent memory as a write loop: what should persist, who can correct it, and how it changes future work.", "answer": [ "Agent memory is not only retrieval; it is a governed write loop that decides what survives a task.", "Useful memory needs correction-time updates, provenance, decay, and behavioral evaluation.", "The failure mode is treating memory as a passive store while the actual system behavior drifts." ], "search_intents": [ "agent memory systems", "LLM memory evaluation", "company memory AI agents" ], "status": "featured", "last_updated": "2026-07-21", "counts": { "total": 13, "signals": 9, "patterns": 1, "posts": 1, "atlas": 2, "sources": 19 }, "routes": { "html": "https://newruntime.com/topics/agent-memory/", "markdown": "https://newruntime.com/topics/agent-memory.md", "json": "https://newruntime.com/topics/agent-memory.json" }, "top_sources": [ "https://aclanthology.org/2024.acl-long.747", "https://careersatdoordash.com/blog/doordash-unified-consumer-memory-for-personalization-at-scale", "https://developers.openai.com/cookbook/examples/agents_sdk/building_reliable_agents_memory_compaction", "https://github.com/AI-Builder-Club/skills", "https://github.com/cocoindex-io/cocoindex", "https://github.com/langchain-ai/deepagents", "https://github.com/langchain-ai/openwiki", "https://github.com/openai/openai-cookbook/blob/main/examples/agents_sdk/building_reliable_agents_memory_compaction.ipynb", "https://github.com/OpenDataBox/MemoryData", "https://github.com/rohitg00/agentmemory", "https://github.com/tinyhumansai/openhuman", "https://github.com/topoteretes/cognee" ], "patterns": [ { "kind": "Pattern", "slug": "company-memory-needs-write-loops", "title": "Company memory needs write and correction loops", "description": "Company memory becomes useful when real work continuously updates a governed knowledge layer, not when an agent merely searches a larger document pile.", "date": "2026-07-16", "topics": [ "company-memory", "agent-memory", "knowledge-systems" ], "source_count": 5, "metric": "high", "url": "https://newruntime.com/patterns/company-memory-needs-write-loops/" } ], "field_notes": [ { "kind": "Field Note", "slug": "cognee-agent-memory-platform", "title": "Cognee Packages Agent Memory as a Self-Hosted Knowledge Graph", "description": "Cognee matters less as another memory SDK and more as an attempt to package persistent agent memory through ingestion, graph/vector search, ontology, and self-hosting.", "date": "2026-07-21", "topics": [ "agent-memory", "knowledge-graph", "open-source" ], "source_count": 1, "url": "https://newruntime.com/posts/cognee-agent-memory-platform/" } ], "raw_signals": [ { "kind": "Raw Signal", "slug": "agent-memory-evaluated-as-system-behavior", "title": "Agent memory is being evaluated as system behavior", "description": "MemoryData evaluates what an agent stores, retrieves, updates, and forgets across a sequence instead of grading one final answer.", "date": "2026-07-06", "topics": [ "agent-memory", "evals", "knowledge-systems" ], "source_count": 2, "metric": "structural", "url": "https://newruntime.com/signals/agent-memory-evaluated-as-system-behavior/" }, { "kind": "Raw Signal", "slug": "codebase-memory-combines-search-with-a-repository-map", "title": "Codebase Memory Combines Search with a Repository Map", "description": "A code retrieval tool adds structural repository context to text search so an agent can locate relevant code without loading the whole project into context.", "date": "2026-07-03", "topics": [ "code-search", "agent-memory", "context-engineering" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/codebase-memory-combines-search-with-a-repository-map/" }, { "kind": "Raw Signal", "slug": "agent-memory-needs-its-own-review-loop", "title": "Agent Memory Needs Its Own Review Loop", "description": "Persistent memory can amplify mistakes across future tasks, so edits, provenance, and correction paths need explicit operator control.", "date": "2026-07-02", "topics": [ "agent-memory", "review", "safety" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/agent-memory-needs-its-own-review-loop/" }, { "kind": "Raw Signal", "slug": "deep-agents-plan-files-memory-subagents", "title": "Deep Agents combines plans, files, memory, and subagents", "description": "The open-source Deep Agents stack packages common long-horizon primitives on top of LangGraph and LangChain.", "date": "2026-07-01", "topics": [ "harness-engineering", "orchestration", "agent-memory" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/deep-agents-plan-files-memory-subagents/" }, { "kind": "Raw Signal", "slug": "doordash-memory-is-versioned-personalization-layer", "title": "DoorDash makes memory a versioned personalization layer", "description": "DoorDash builds structured memory blocks from behavior and stores model, schema, timestamp, prompt, and response lineage for each component.", "date": "2026-06-24", "topics": [ "agent-memory", "personalization", "data-lineage" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/doordash-memory-is-versioned-personalization-layer/" }, { "kind": "Raw Signal", "slug": "openhuman-builds-a-personal-harness-around-local-memory", "title": "OpenHuman Builds a Personal Harness Around Local Memory", "description": "OpenHuman combines personal data, memory, tools, and an operator layer into a durable system instead of another isolated assistant chat.", "date": "2026-05-24", "topics": [ "personal-agents", "local-first", "agent-memory" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/openhuman-builds-a-personal-harness-around-local-memory/" }, { "kind": "Raw Signal", "slug": "agentmemory-shares-state-across-codex-claude-code-and-cursor", "title": "AgentMemory Shares State Across Codex, Claude Code, and Cursor", "description": "A common memory layer reduces the cost of switching coding clients and moves context ownership outside any one model interface.", "date": "2026-05-19", "topics": [ "agent-memory", "coding-agents", "interoperability" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/agentmemory-shares-state-across-codex-claude-code-and-cursor/" }, { "kind": "Raw Signal", "slug": "openai-shows-memory-compaction-for-long-running-agents", "title": "OpenAI Shows Memory Compaction for Long-Running Agents", "description": "The Agents SDK pattern treats compaction and state transfer as explicit reliability mechanisms for work that outlives one context window.", "date": "2026-05-07", "topics": [ "agent-memory", "compaction", "long-running-agents" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/openai-shows-memory-compaction-for-long-running-agents/" }, { "kind": "Raw Signal", "slug": "cocoindex-keeps-agent-context-fresh-without-full-reindexing", "title": "CocoIndex Keeps Agent Context Fresh Without Full Reindexing", "description": "CocoIndex turns source changes into incremental dataflow updates for live agent context and code-aware retrieval.", "date": "2026-05-06", "topics": [ "agent-memory", "retrieval", "data-infrastructure" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/cocoindex-keeps-agent-context-fresh-without-full-reindexing/" } ], "atlas_records": [ { "kind": "Shift", "slug": "shared-context-infrastructure", "title": "Client-local memory -> shared context infrastructure", "description": "Context is shifting from private memory inside one AI client into a governed layer between agents, IDEs, documents, and tools.", "date": "2026-07-21", "topics": [ "context-engineering", "agent-memory", "mcp" ], "source_count": 2, "metric": "emerging", "url": "https://newruntime.com/shifts/shared-context-infrastructure/" }, { "kind": "Graveyard", "slug": "manual-context-handoff", "title": "Manual context handoff between AI clients", "description": "The ritual of retelling one AI client what another already understood becomes optional when context moves into a shared layer.", "date": "2026-07-21", "topics": [ "context-engineering", "agent-memory" ], "source_count": 1, "metric": "challenged", "url": "https://newruntime.com/graveyard/manual-context-handoff/" } ] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "skills", "title": "Agent skills - New Runtime", "description": "Portable skill files, behavioral contracts, tool instructions, and evals that make agent capability reusable.", "answer": [ "Agent skills package repeatable behavior outside the current chat.", "The important boundary is whether the skill changes task outcomes, not whether the instruction prose looks good.", "Skills become infrastructure when they are versioned, evaluated, and shared across agents." ], "search_intents": [ "agent skills", "AI agent skill files", "reusable agent capabilities" ], "status": "featured", "last_updated": "2026-07-04", "counts": { "total": 13, "signals": 13, "patterns": 0, "posts": 0, "atlas": 0, "sources": 22 }, "routes": { "html": "https://newruntime.com/topics/skills/", "markdown": "https://newruntime.com/topics/skills.md", "json": "https://newruntime.com/topics/skills.json" }, "top_sources": [ "https://claude.com/blog/steering-claude-code-skills-hooks-rules-subagents-and-more", "https://cursor.com/marketplace/cursor/cursor-team-kit", "https://developers.openai.com/codex/record-and-replay", "https://github.com/anthropics/claude-for-legal", "https://github.com/emilkowalski/skills", "https://github.com/mvanhorn/last30days-skil", "https://github.com/mvanhorn/last30days-skill.git", "https://github.com/NousResearch/hermes-agent/blob/main/optional-skills/creative/creative-ideation/SKILL.md", "https://github.com/nvidia/skillspector", "https://github.com/openai/skills/tree/main/skills/.curated/gh-address-comments", "https://github.com/senlindesign/taste-skill", "https://huggingface.co/" ], "patterns": [], "field_notes": [], "raw_signals": [ { "kind": "Raw Signal", "slug": "animation-vocabulary-turns-motion-references-into-an-agent-skill", "title": "Animation Vocabulary Turns Motion References into an Agent Skill", "description": "A compact design skill maps visual intent to named motion patterns, giving interface agents a reusable language for implementing animation decisions.", "date": "2026-07-04", "topics": [ "design-systems", "animation", "skills" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/animation-vocabulary-turns-motion-references-into-an-agent-skill/" }, { "kind": "Raw Signal", "slug": "a-github-skill-can-resolve-review-comments-as-a-batch", "title": "A GitHub Skill Can Resolve Review Comments as a Batch", "description": "A reusable Codex workflow gathers pull-request feedback, plans changes, applies fixes, and reports resolution state as one auditable review task.", "date": "2026-07-03", "topics": [ "code-review", "skills", "github" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/a-github-skill-can-resolve-review-comments-as-a-batch/" }, { "kind": "Raw Signal", "slug": "creative-ideation-becomes-a-routable-agent-skill", "title": "Creative Ideation Becomes a Routable Agent Skill", "description": "A packaged workflow decomposes creative exploration into reference gathering, divergence, critique, and selection instead of asking a model to be creative in one step.", "date": "2026-06-27", "topics": [ "creative-work", "skills", "design" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/creative-ideation-becomes-a-routable-agent-skill/" }, { "kind": "Raw Signal", "slug": "agentic-resource-discovery-indexes-skills-mcp-and-agents", "title": "Agentic Resource Discovery Indexes Skills, MCP, and Agents", "description": "Hugging Face treats agent capabilities as searchable resources, creating a discovery layer for tools that can be loaded at runtime.", "date": "2026-06-24", "topics": [ "agent-discovery", "mcp", "skills" ], "source_count": 3, "metric": "notable", "url": "https://newruntime.com/signals/agentic-resource-discovery-indexes-skills-mcp-and-agents/" }, { "kind": "Raw Signal", "slug": "hermes-turns-solved-tasks-into-new-skills", "title": "Hermes Turns Solved Tasks into New Skills", "description": "An agent runtime can preserve successful procedures as reusable capability modules, converting execution traces into a compounding operational library.", "date": "2026-06-24", "topics": [ "hermes", "skills", "agent-learning" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/hermes-turns-solved-tasks-into-new-skills/" }, { "kind": "Raw Signal", "slug": "claude-md-shrinks-as-skills-hooks-and-subagents-mature", "title": "CLAUDE.md Shrinks as Skills, Hooks, and Subagents Mature", "description": "Agent behavior moves out of one giant context file into scoped mechanisms that load only for the relevant task and lifecycle stage.", "date": "2026-06-23", "topics": [ "claude-code", "skills", "context-engineering" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/claude-md-shrinks-as-skills-hooks-and-subagents-mature/" }, { "kind": "Raw Signal", "slug": "codex-turns-a-recorded-mac-workflow-into-a-reusable-skill", "title": "Codex Turns a Recorded Mac Workflow into a Reusable Skill", "description": "Record and Replay captures a human procedure, converts it into agent instructions, and preserves the successful interaction as reusable behavior.", "date": "2026-06-22", "topics": [ "codex", "skills", "computer-use" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/codex-turns-a-recorded-mac-workflow-into-a-reusable-skill/" }, { "kind": "Raw Signal", "slug": "taste-lab-uses-a-design-system-as-a-reasoning-brief", "title": "Taste Lab Uses a Design System as a Reasoning Brief", "description": "Taste Lab packages aesthetic decisions and implementation rules as agent-readable context rather than relying on vague style prompts.", "date": "2026-06-19", "topics": [ "design-systems", "skills", "frontend" ], "source_count": 3, "metric": "notable", "url": "https://newruntime.com/signals/taste-lab-uses-a-design-system-as-a-reasoning-brief/" }, { "kind": "Raw Signal", "slug": "last30days-packages-social-research-as-an-agent-skill", "title": "Last30Days Packages Social Research as an Agent Skill", "description": "A reusable research workflow gathers recent public discussion, filters repetition, and produces a time-bounded evidence set for another agent or analyst.", "date": "2026-06-18", "topics": [ "research-agents", "skills", "social-data" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/last30days-packages-social-research-as-an-agent-skill/" }, { "kind": "Raw Signal", "slug": "skillspector-adds-a-security-gate-for-agent-skills", "title": "SkillSpector Adds a Security Gate for Agent Skills", "description": "NVIDIA's scanner treats installable agent instructions as executable supply-chain artifacts that require inspection before use.", "date": "2026-06-02", "topics": [ "agent-security", "skills", "supply-chain" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/skillspector-adds-a-security-gate-for-agent-skills/" }, { "kind": "Raw Signal", "slug": "claude-for-legal-packages-a-legal-team-as-plugins", "title": "Claude for Legal Packages a Legal Team as Plugins", "description": "Anthropic's legal repository expresses domain knowledge, procedures, and tools as installable modules for repeatable professional work.", "date": "2026-05-20", "topics": [ "legal-agents", "skills", "enterprise-ai" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/claude-for-legal-packages-a-legal-team-as-plugins/" }, { "kind": "Raw Signal", "slug": "skill-md-acts-as-a-behavior-loader-not-a-better-prompt", "title": "SKILL.md Acts as a Behavior Loader, Not a Better Prompt", "description": "Skills package procedures, tools, and progressive context so an agent can load behavior only when the task requires it.", "date": "2026-05-07", "topics": [ "skills", "context-engineering", "agent-harness" ], "source_count": 3, "metric": "notable", "url": "https://newruntime.com/signals/skill-md-acts-as-a-behavior-loader-not-a-better-prompt/" }, { "kind": "Raw Signal", "slug": "cursor-ships-its-development-procedures-as-agent-skills", "title": "Cursor Ships Its Development Procedures as Agent Skills", "description": "Cursor Team Kit packages verification, CI repair, review preparation, and code cleanup as reusable agent behavior.", "date": "2026-05-06", "topics": [ "skills", "coding-agents", "verification" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/cursor-ships-its-development-procedures-as-agent-skills/" } ], "atlas_records": [] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "model-routing", "title": "Model routing - New Runtime", "description": "How teams route work across model families, effort levels, cheap executors, judges, and local/runtime constraints.", "answer": [ "Model routing treats models as swappable execution resources rather than a single product choice.", "The pattern is to match task risk, latency, cost, and verification needs to the right model or model sequence.", "This makes the harness more durable than any one frontier model release." ], "search_intents": [ "model routing", "multi model AI systems", "LLM routing architecture" ], "status": "featured", "last_updated": "2026-07-16", "counts": { "total": 13, "signals": 11, "patterns": 2, "posts": 0, "atlas": 0, "sources": 25 }, "routes": { "html": "https://newruntime.com/topics/model-routing/", "markdown": "https://newruntime.com/topics/model-routing.md", "json": "https://newruntime.com/topics/model-routing.json" }, "top_sources": [ "https://claude.com/blog/claude-model-and-effort-level-in-claude-code", "https://code.claude.com/docs/en/model-config", "https://cognition.com/blog/making-fable-cheaper-than-opus", "https://databricks.com/blog/benchmarking-coding-agents-databricks-multi-million-line-codebase", "https://docs.inference.net/introduction", "https://engineering.ramp.com/post/ai-spend-value", "https://github.com/langchain-ai/deepagents", "https://github.com/NousResearch/hermes-agent", "https://github.com/rivet-dev/agentos", "https://github.com/shadcn/improve", "https://github.com/shadcn/improve/blob/main/skills/improve/SKILL.md", "https://github.com/steipete/agent-scripts/blob/main/skills/codex-first/SKILL.md" ], "patterns": [ { "kind": "Pattern", "slug": "harness-architecture-outlives-model-choice", "title": "Harness architecture outlives model choice", "description": "For production agents, the harness is becoming a more durable product boundary than the identity of the model running inside it.", "date": "2026-07-16", "topics": [ "harness-engineering", "agent-runtime", "model-routing" ], "source_count": 5, "metric": "high", "url": "https://newruntime.com/patterns/harness-architecture-outlives-model-choice/" }, { "kind": "Pattern", "slug": "agent-economics-moves-to-completed-work", "title": "Agent economics moves to completed work", "description": "The economically meaningful unit for agent systems is becoming cost per verified completed task rather than cost per token or model call.", "date": "2026-07-15", "topics": [ "agent-economics", "model-routing", "product-metrics" ], "source_count": 5, "metric": "high", "url": "https://newruntime.com/patterns/agent-economics-moves-to-completed-work/" } ], "field_notes": [], "raw_signals": [ { "kind": "Raw Signal", "slug": "opencode-model-swappable-agent-harness", "title": "OpenCode makes the model a swappable dependency", "description": "OpenCode keeps rules, skills, permissions, sessions, and tools stable while the underlying model changes.", "date": "2026-07-16", "topics": [ "coding-agents", "harness-engineering", "model-routing", "developer-tools" ], "source_count": 4, "metric": "structural", "url": "https://newruntime.com/signals/opencode-model-swappable-agent-harness/" }, { "kind": "Raw Signal", "slug": "fable-cost-per-completed-task", "title": "A pricier model can be cheaper per completed task", "description": "Cognition reports that Fable 5 completed coding work with fewer steps and output tokens than its previous lead model.", "date": "2026-07-15", "topics": [ "agent-economics", "coding-agents", "model-routing" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/fable-cost-per-completed-task/" }, { "kind": "Raw Signal", "slug": "claude-code-model-and-effort-are-separate-controls", "title": "Claude Code separates model choice from effort", "description": "Anthropic exposes model selection and effort level as different controls for capability, token use, latency, and persistence.", "date": "2026-07-13", "topics": [ "coding-agents", "model-routing", "agent-economics" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/claude-code-model-and-effort-are-separate-controls/" }, { "kind": "Raw Signal", "slug": "improve-routes-architect-and-executor-models", "title": "Improve routes architecture and execution to different models", "description": "The Improve skill audits a repository with a stronger planning model and delegates bounded fixes to cheaper executors.", "date": "2026-07-12", "topics": [ "agent-skills", "model-routing", "code-review" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/improve-routes-architect-and-executor-models/" }, { "kind": "Raw Signal", "slug": "advisor-model-guides-cheaper-executor", "title": "An advisor model can guide a cheaper executor", "description": "The advisor-tool pattern lets a fast executor request bounded analysis from a stronger model while keeping control of the task loop.", "date": "2026-07-09", "topics": [ "model-routing", "orchestration", "agent-economics" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/advisor-model-guides-cheaper-executor/" }, { "kind": "Raw Signal", "slug": "inference-gateway-makes-model-switching-operational", "title": "An inference gateway makes model switching operational", "description": "Inference.net exposes model routing through a stable gateway so clients can change providers without rewriting every integration.", "date": "2026-07-02", "topics": [ "model-routing", "inference", "agent-infrastructure" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/inference-gateway-makes-model-switching-operational/" }, { "kind": "Raw Signal", "slug": "coinbase-reports-lower-ai-spend-as-token-use-rises", "title": "Coinbase Reports Lower AI Spend as Token Use Rises", "description": "A production account suggests routing, caching, model choice, and workflow design can reduce unit cost even while aggregate agent usage expands.", "date": "2026-06-28", "topics": [ "ai-economics", "model-routing", "enterprise-ai" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/coinbase-reports-lower-ai-spend-as-token-use-rises/" }, { "kind": "Raw Signal", "slug": "glm-5-2-changes-the-economics-of-claude-code-workflows", "title": "GLM-5.2 Changes the Economics of Claude Code Workflows", "description": "A capable open model routed through compatible infrastructure can replace premium inference for parts of a coding-agent workload.", "date": "2026-06-27", "topics": [ "open-models", "model-routing", "coding-agents" ], "source_count": 4, "metric": "notable", "url": "https://newruntime.com/signals/glm-5-2-changes-the-economics-of-claude-code-workflows/" }, { "kind": "Raw Signal", "slug": "sakana-fugu-presents-multiple-agents-as-one-model-endpoint", "title": "Sakana Fugu Presents Multiple Agents as One Model Endpoint", "description": "Fugu hides a multi-agent debate and synthesis system behind a model-like interface, making orchestration an implementation detail of inference.", "date": "2026-06-25", "topics": [ "multi-agent", "model-routing", "inference" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/sakana-fugu-presents-multiple-agents-as-one-model-endpoint/" }, { "kind": "Raw Signal", "slug": "fusion-api-uses-multiple-models-and-a-judge-for-one-answer", "title": "Fusion API Uses Multiple Models and a Judge for One Answer", "description": "A single endpoint can route a question to several models and synthesize their outputs, moving model selection and comparison behind an orchestration layer.", "date": "2026-06-17", "topics": [ "model-routing", "multi-model", "evaluation" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/fusion-api-uses-multiple-models-and-a-judge-for-one-answer/" }, { "kind": "Raw Signal", "slug": "vercel-treats-production-ai-as-multi-model-infrastructure", "title": "Vercel Treats Production AI as Multi-Model Infrastructure", "description": "Vercel's gateway view assumes model providers are routable dependencies with fallback, observability, and policy around them.", "date": "2026-05-15", "topics": [ "model-routing", "ai-gateway", "production-ai" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/vercel-treats-production-ai-as-multi-model-infrastructure/" } ], "atlas_records": [] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "orchestration", "title": "Orchestration - New Runtime", "description": "How agent work is split into queues, roles, subagents, async loops, handoffs, and operator-visible state.", "answer": [ "Agent orchestration is the control layer around multiple autonomous or semi-autonomous work loops.", "It becomes necessary when one prompt cannot hold planning, execution, review, memory, and user coordination at once.", "The visible product shift is from a chat box to an operator surface for supervising work." ], "search_intents": [ "agent orchestration", "multi agent workflow", "AI agent control plane" ], "status": "featured", "last_updated": "2026-07-15", "counts": { "total": 12, "signals": 12, "patterns": 0, "posts": 0, "atlas": 0, "sources": 17 }, "routes": { "html": "https://newruntime.com/topics/orchestration/", "markdown": "https://newruntime.com/topics/orchestration.md", "json": "https://newruntime.com/topics/orchestration.json" }, "top_sources": [ "https://academy.langchain.com/courses/foundation-introduction-to-deepagents", "https://arxiv.org/abs/2607.01942", "https://claude.com/blog/new-in-claude-managed-agents", "https://github.com/anthropics/claude-cookbooks", "https://github.com/langchain-ai/deepagents", "https://github.com/ogulcancelik/herdr", "https://github.com/steipete/agent-scripts/blob/main/skills/codex-first/SKILL.md", "https://herdr.dev/install.sh", "https://hermes-agent.nousresearch.com/docs/user-guide/features/kanban", "https://hermes-agent.nousresearch.com/docs/user-guide/features/kanban-tutorial", "https://mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/from-campaigns-to-continuous-growth-ai-capabilities-shaping-marketing", "https://platform.claude.com/cookbook/patterns-agents-async-multi-agent-orchestration" ], "patterns": [], "field_notes": [], "raw_signals": [ { "kind": "Raw Signal", "slug": "atomic-task-graphs-outperform-react-loops", "title": "Atomic task graphs can let smaller models beat larger loops", "description": "An Atomic Task Graph decomposes work into validated dependencies, parallel branches, and locally recoverable failures.", "date": "2026-07-15", "topics": [ "agent-architecture", "orchestration", "task-graphs" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/atomic-task-graphs-outperform-react-loops/" }, { "kind": "Raw Signal", "slug": "deep-agents-package-long-running-harness", "title": "Deep Agents packages a harness for long-running tasks", "description": "LangChain's Deep Agents combines planning, files, memory, subagents, middleware, and backends for multi-step work.", "date": "2026-07-10", "topics": [ "harness-engineering", "agent-loops", "orchestration" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/deep-agents-package-long-running-harness/" }, { "kind": "Raw Signal", "slug": "advisor-model-guides-cheaper-executor", "title": "An advisor model can guide a cheaper executor", "description": "The advisor-tool pattern lets a fast executor request bounded analysis from a stronger model while keeping control of the task loop.", "date": "2026-07-09", "topics": [ "model-routing", "orchestration", "agent-economics" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/advisor-model-guides-cheaper-executor/" }, { "kind": "Raw Signal", "slug": "herdr-adds-agent-awareness-to-terminal-multiplexing", "title": "Herdr Adds Agent Awareness to Terminal Multiplexing", "description": "Herdr organizes parallel coding-agent sessions around projects, process state, and operator navigation rather than presenting them as undifferentiated terminal panes.", "date": "2026-07-04", "topics": [ "coding-agents", "orchestration", "terminal" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/herdr-adds-agent-awareness-to-terminal-multiplexing/" }, { "kind": "Raw Signal", "slug": "the-future-interface-is-an-agent-manager-not-twenty-terminals", "title": "The Future Interface Is an Agent Manager, Not Twenty Terminals", "description": "As parallelism grows, users need one surface for goals, status, handoffs, exceptions, and review rather than more chat windows.", "date": "2026-07-03", "topics": [ "operator-ux", "multi-agent", "orchestration" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/the-future-interface-is-an-agent-manager-not-twenty-terminals/" }, { "kind": "Raw Signal", "slug": "ai-reorganizes-marketing-around-search-and-orchestration", "title": "AI Reorganizes Marketing Around Search and Orchestration", "description": "McKinsey frames marketing work as a coordinated system of human judgment, AI search visibility, generated content, and agent-mediated customer journeys.", "date": "2026-07-02", "topics": [ "marketing", "agentic-commerce", "orchestration" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/ai-reorganizes-marketing-around-search-and-orchestration/" }, { "kind": "Raw Signal", "slug": "deep-agents-plan-files-memory-subagents", "title": "Deep Agents combines plans, files, memory, and subagents", "description": "The open-source Deep Agents stack packages common long-horizon primitives on top of LangGraph and LangChain.", "date": "2026-07-01", "topics": [ "harness-engineering", "orchestration", "agent-memory" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/deep-agents-plan-files-memory-subagents/" }, { "kind": "Raw Signal", "slug": "warp-reframes-product-engineers-as-factory-engineers", "title": "Warp Reframes Product Engineers as Factory Engineers", "description": "An AI-native engineering organization invests more effort in task decomposition, agent infrastructure, review systems, and throughput management than in direct implementation.", "date": "2026-06-30", "topics": [ "engineering-org", "coding-agents", "orchestration" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/warp-reframes-product-engineers-as-factory-engineers/" }, { "kind": "Raw Signal", "slug": "cursor-is-expanding-from-ide-to-agent-factory", "title": "Cursor Is Expanding from IDE to Agent Factory", "description": "Cursor's workflow increasingly centers on parallel autonomous tasks, remote execution, review, and orchestration rather than one developer editing one file at a time.", "date": "2026-06-27", "topics": [ "cursor", "coding-agents", "orchestration" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/cursor-is-expanding-from-ide-to-agent-factory/" }, { "kind": "Raw Signal", "slug": "anthropic-shows-two-async-orchestration-patterns", "title": "Anthropic Shows Two Async Orchestration Patterns", "description": "Parallel agents can be coordinated through explicit state and asynchronous handoffs instead of sharing one overloaded conversation.", "date": "2026-06-16", "topics": [ "orchestration", "async-agents", "multi-agent" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/anthropic-shows-two-async-orchestration-patterns/" }, { "kind": "Raw Signal", "slug": "claude-managed-agents-add-outcomes-and-multi-agent-orchestration", "title": "Claude Managed Agents Add Outcomes and Multi-Agent Orchestration", "description": "Anthropic's managed-agent layer adds long-running outcomes, orchestration, and background work around the model call.", "date": "2026-05-07", "topics": [ "managed-agents", "orchestration", "agent-runtime" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/claude-managed-agents-add-outcomes-and-multi-agent-orchestration/" }, { "kind": "Raw Signal", "slug": "hermes-profiles-need-a-shared-work-layer-to-become-a-crew", "title": "Hermes Profiles Need a Shared Work Layer to Become a Crew", "description": "Multiple specialist profiles become a system only when they share tasks, dependencies, handoffs, and a persistent operational board.", "date": "2026-05-07", "topics": [ "multi-agent", "orchestration", "hermes" ], "source_count": 3, "metric": "notable", "url": "https://newruntime.com/signals/hermes-profiles-need-a-shared-work-layer-to-become-a-crew/" } ], "atlas_records": [] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "agent-runtime", "title": "Agent runtime - New Runtime", "description": "The runtime layer around agents: sandboxes, tools, hosted execution, permissions, state, and deployment surfaces.", "answer": [ "The agent runtime is the system that lets models safely use tools, files, browsers, terminals, and external services.", "As agents take real actions, runtime concerns become product concerns: isolation, permissions, logging, replay, and rollback.", "This is where demos turn into repeatable infrastructure." ], "search_intents": [ "agent runtime", "AI agent infrastructure", "sandboxed agent execution" ], "status": "featured", "last_updated": "2026-07-22", "counts": { "total": 11, "signals": 9, "patterns": 1, "posts": 1, "atlas": 0, "sources": 23 }, "routes": { "html": "https://newruntime.com/topics/agent-runtime/", "markdown": "https://newruntime.com/topics/agent-runtime.md", "json": "https://newruntime.com/topics/agent-runtime.json" }, "top_sources": [ "https://agentos-sdk.dev/", "https://claude.com/blog/new-in-claude-managed-agents", "https://cline.bot/blog/introducing-cline-sdk-the-upgraded-agent-runtime", "https://developers.cloudflare.com/sandbox/tutorials/devin-outposts/", "https://docs.inference.net/introduction", "https://github.com/anthropics/claude-cookbooks/blob/main/claude_agent_sdk/07_Hosting_the_agent.ipynb", "https://github.com/bytedance/deer-flow", "https://github.com/cline/cline/tree/main/sdk/apps/examples", "https://github.com/langchain-ai/deepagents", "https://github.com/NousResearch/hermes-agent", "https://github.com/rivet-dev/agentos", "https://langchain.com/blog/give-your-ai-agent-its-own-computer" ], "patterns": [ { "kind": "Pattern", "slug": "harness-architecture-outlives-model-choice", "title": "Harness architecture outlives model choice", "description": "For production agents, the harness is becoming a more durable product boundary than the identity of the model running inside it.", "date": "2026-07-16", "topics": [ "harness-engineering", "agent-runtime", "model-routing" ], "source_count": 5, "metric": "high", "url": "https://newruntime.com/patterns/harness-architecture-outlives-model-choice/" } ], "field_notes": [ { "kind": "Field Note", "slug": "devin-outposts-agent-execution-plane", "title": "Devin Outposts Splits the Agent Brain from the Execution Plane", "description": "Devin Outposts moves command execution, repository access, and sandbox lifecycle into customer-controlled infrastructure while the agent loop remains in Cognition's cloud.", "date": "2026-07-22", "topics": [ "coding-agents", "agent-runtime", "infrastructure" ], "source_count": 4, "url": "https://newruntime.com/posts/devin-outposts-agent-execution-plane/" } ], "raw_signals": [ { "kind": "Raw Signal", "slug": "render-packages-deployment-primitives-for-agent-workloads", "title": "Render Packages Deployment Primitives for Agent Workloads", "description": "Render combines services, workers, databases, cron jobs, previews, and infrastructure definitions into a deployment surface suitable for long-running agent systems.", "date": "2026-07-11", "topics": [ "deployment", "agent-runtime", "infrastructure" ], "source_count": 3, "metric": "notable", "url": "https://newruntime.com/signals/render-packages-deployment-primitives-for-agent-workloads/" }, { "kind": "Raw Signal", "slug": "agentos-lightweight-runtime-for-coding-agents", "title": "agentOS packages a lightweight runtime for coding agents", "description": "Rivet's agentOS combines isolated execution, durable workflows, resource limits, and host-side secrets in a WebAssembly and Rust runtime.", "date": "2026-06-29", "topics": [ "agent-runtime", "sandboxes", "coding-agents" ], "source_count": 3, "metric": "notable", "url": "https://newruntime.com/signals/agentos-lightweight-runtime-for-coding-agents/" }, { "kind": "Raw Signal", "slug": "deerflow-2-0-packages-an-open-runtime-for-autonomous-agents", "title": "DeerFlow 2.0 Packages an Open Runtime for Autonomous Agents", "description": "ByteDance's open runtime combines planning, subagents, files, tools, and long-running execution into a deployable system.", "date": "2026-06-28", "topics": [ "open-source", "multi-agent", "agent-runtime" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/deerflow-2-0-packages-an-open-runtime-for-autonomous-agents/" }, { "kind": "Raw Signal", "slug": "ai-sdk-7-grows-into-an-agent-runtime", "title": "AI SDK 7 Grows into an Agent Runtime", "description": "Vercel expands a model integration library with tool loops, state, transport, and UI primitives needed for production agent applications.", "date": "2026-06-27", "topics": [ "ai-sdk", "agent-runtime", "vercel" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/ai-sdk-7-grows-into-an-agent-runtime/" }, { "kind": "Raw Signal", "slug": "the-computer-becomes-a-runtime-assigned-to-the-agent", "title": "The Computer Becomes a Runtime Assigned to the Agent", "description": "Giving each agent an isolated machine turns browser, shell, files, and installed software into controlled execution infrastructure.", "date": "2026-06-19", "topics": [ "sandboxes", "agent-runtime", "computer-use" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/the-computer-becomes-a-runtime-assigned-to-the-agent/" }, { "kind": "Raw Signal", "slug": "tokens-stay-the-cost-while-runtime-becomes-the-product", "title": "Tokens Stay the Cost While Runtime Becomes the Product", "description": "The differentiated layer shifts toward execution, tools, policy, state, and user workflow while model inference remains metered input.", "date": "2026-06-18", "topics": [ "agent-runtime", "product-strategy", "ai-economics" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/tokens-stay-the-cost-while-runtime-becomes-the-product/" }, { "kind": "Raw Signal", "slug": "the-agent-becomes-a-separate-hosting-layer", "title": "The Agent Becomes a Separate Hosting Layer", "description": "Anthropic's Agent SDK hosting pattern separates model calls from a persistent process that owns tools, state, execution, and recovery.", "date": "2026-06-16", "topics": [ "agent-runtime", "hosting", "agent-sdk" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/the-agent-becomes-a-separate-hosting-layer/" }, { "kind": "Raw Signal", "slug": "cline-sdk-exposes-an-open-harness-for-coding-agents", "title": "Cline SDK Exposes an Open Harness for Coding Agents", "description": "Cline separates the reusable agent runtime from its interface, making tool execution and coding workflows embeddable in other products.", "date": "2026-05-14", "topics": [ "coding-agents", "sdk", "agent-runtime" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/cline-sdk-exposes-an-open-harness-for-coding-agents/" }, { "kind": "Raw Signal", "slug": "claude-managed-agents-add-outcomes-and-multi-agent-orchestration", "title": "Claude Managed Agents Add Outcomes and Multi-Agent Orchestration", "description": "Anthropic's managed-agent layer adds long-running outcomes, orchestration, and background work around the model call.", "date": "2026-05-07", "topics": [ "managed-agents", "orchestration", "agent-runtime" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/claude-managed-agents-add-outcomes-and-multi-agent-orchestration/" } ], "atlas_records": [] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "verification", "title": "Verification - New Runtime", "description": "Signals and patterns about verification as the limiting factor for agent-produced work and AI-native operations.", "answer": [ "Verification is becoming the scarce resource in agentic work.", "More generation only helps when review, tests, traces, and provenance can keep up.", "The best systems make verification part of the product surface instead of an afterthought." ], "search_intents": [ "AI verification", "agent verification", "AI generated code review" ], "status": "featured", "last_updated": "2026-07-21", "counts": { "total": 10, "signals": 9, "patterns": 0, "posts": 1, "atlas": 0, "sources": 13 }, "routes": { "html": "https://newruntime.com/topics/verification/", "markdown": "https://newruntime.com/topics/verification.md", "json": "https://newruntime.com/topics/verification.json" }, "top_sources": [ "https://agent-cookbook.com/tutorial/ai-20260512163240-54x4", "https://blog.bytebytego.com/p/how-openai-built-its-data-agent", "https://code.claude.com/docs/en/goal", "https://cursor.com/marketplace/cursor/cursor-team-kit", "https://cutlefish.substack.com/p/tbm-427-the-bottleneck-strike-again", "https://developer.chrome.com/docs/devtools/agents", "https://developer.chrome.com/docs/modern-web-guidance", "https://github.com/mattpocock/skills", "https://trycase.dev/docs", "https://www.youtube.com/watch?v=-QFHIoCo-Ko", "https://x.com/ClaudeDevs/status/2074208949205881033", "https://x.com/ericzakariasson/status/2051255780167610522" ], "patterns": [], "field_notes": [ { "kind": "Field Note", "slug": "ai-coding-workflow-verifiable-work", "title": "AI Coding Workflow: From Idea to Verifiable Work", "description": "AI coding works better when checkable artifacts stand between the idea and the code: specs, tickets, TDD, fresh-context review, and manual QA.", "date": "2026-07-21", "topics": [ "coding-agents", "workflow", "verification" ], "source_count": 2, "url": "https://newruntime.com/posts/ai-coding-workflow-verifiable-work/" } ], "raw_signals": [ { "kind": "Raw Signal", "slug": "cheap-code-moves-bottleneck-to-review", "title": "Cheap code moves the engineering bottleneck to review", "description": "An agent-heavy development model treats implementation as abundant while specifications, validation, security, and integration remain scarce.", "date": "2026-07-12", "topics": [ "engineering-management", "coding-agents", "verification" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/cheap-code-moves-bottleneck-to-review/" }, { "kind": "Raw Signal", "slug": "the-future-engineer-is-defined-by-judgment-not-typing-speed", "title": "The Future Engineer Is Defined by Judgment, Not Typing Speed", "description": "As implementation becomes cheaper, system framing, constraint design, verification, and ownership become more valuable than raw code production speed.", "date": "2026-07-12", "topics": [ "engineering-roles", "judgment", "verification" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/the-future-engineer-is-defined-by-judgment-not-typing-speed/" }, { "kind": "Raw Signal", "slug": "trycase-disposable-linux-agent-verification", "title": "TryCase gives coding agents disposable Linux verification", "description": "TryCase creates short-lived Linux environments where agents can run generated code and tests away from the developer's machine.", "date": "2026-07-10", "topics": [ "sandboxes", "coding-agents", "verification" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/trycase-disposable-linux-agent-verification/" }, { "kind": "Raw Signal", "slug": "loop-engineering-treats-the-agent-as-a-process-not-a-prompt", "title": "Loop Engineering Treats the Agent as a Process, Not a Prompt", "description": "The useful unit of agent design becomes a repeated cycle with state, tools, checks, and stopping conditions rather than a single carefully worded instruction.", "date": "2026-07-08", "topics": [ "loop-engineering", "agents", "verification" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/loop-engineering-treats-the-agent-as-a-process-not-a-prompt/" }, { "kind": "Raw Signal", "slug": "ai-moves-the-bottleneck-across-the-whole-product-system", "title": "AI Moves the Bottleneck Across the Whole Product System", "description": "Faster code generation exposes constraints in product decisions, review, integration, and distribution instead of eliminating the delivery bottleneck.", "date": "2026-07-02", "topics": [ "product-development", "bottlenecks", "verification" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/ai-moves-the-bottleneck-across-the-whole-product-system/" }, { "kind": "Raw Signal", "slug": "openai-built-a-data-agent-across-ninety-thousand-tables", "title": "OpenAI Built a Data Agent Across Ninety Thousand Tables", "description": "A large internal data environment requires metadata discovery, permission-aware retrieval, query validation, and feedback loops beyond a generic text-to-SQL prompt.", "date": "2026-06-23", "topics": [ "data-agents", "enterprise-data", "verification" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/openai-built-a-data-agent-across-ninety-thousand-tables/" }, { "kind": "Raw Signal", "slug": "chrome-turns-web-development-into-a-verifiable-agent-loop", "title": "Chrome Turns Web Development into a Verifiable Agent Loop", "description": "Browser guidance, inspection, interaction, and screenshots make frontend work observable enough for agents to test rather than merely generate.", "date": "2026-05-20", "topics": [ "browser-tools", "frontend", "verification" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/chrome-turns-web-development-into-a-verifiable-agent-loop/" }, { "kind": "Raw Signal", "slug": "claude-code-goals-turn-completion-into-a-verifiable-contract", "title": "Claude Code Goals Turn Completion into a Verifiable Contract", "description": "A goal binds autonomous work to an explicit outcome and stopping rule instead of relying on repeated keep-going prompts.", "date": "2026-05-14", "topics": [ "goals", "long-running-agents", "verification" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/claude-code-goals-turn-completion-into-a-verifiable-contract/" }, { "kind": "Raw Signal", "slug": "cursor-ships-its-development-procedures-as-agent-skills", "title": "Cursor Ships Its Development Procedures as Agent Skills", "description": "Cursor Team Kit packages verification, CI repair, review preparation, and code cleanup as reusable agent behavior.", "date": "2026-05-06", "topics": [ "skills", "coding-agents", "verification" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/cursor-ships-its-development-procedures-as-agent-skills/" } ], "atlas_records": [] }, { "schema_version": "newruntime-topic-hub-v0.1", "type": "topic_hub", "slug": "agent-economics", "title": "Agent economics - New Runtime", "description": "How AI work is priced, measured, routed, and justified by completed task value rather than raw token volume.", "answer": [ "Agent economics moves measurement from token use to useful completed work.", "Cost only makes sense beside success rate, review time, retry loops, latency, and human attention.", "The mature metric is cost per accepted outcome, not cost per generation." ], "search_intents": [ "agent economics", "AI coding cost", "cost per completed AI task" ], "status": "featured", "last_updated": "2026-07-22", "counts": { "total": 8, "signals": 6, "patterns": 1, "posts": 1, "atlas": 0, "sources": 9 }, "routes": { "html": "https://newruntime.com/topics/agent-economics/", "markdown": "https://newruntime.com/topics/agent-economics.md", "json": "https://newruntime.com/topics/agent-economics.json" }, "top_sources": [ "https://anthropic.com/news/claude-sonnet-5", "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/", "https://claude.com/blog/claude-model-and-effort-level-in-claude-code", "https://code.claude.com/docs/en/model-config", "https://cognition.com/blog/making-fable-cheaper-than-opus", "https://databricks.com/blog/benchmarking-coding-agents-databricks-multi-million-line-codebase", "https://engineering.ramp.com/post/ai-spend-value", "https://github.com/steipete/agent-scripts/blob/main/skills/codex-first/SKILL.md", "https://platform.claude.com/docs/en/agents-and-tools/tool-use/advisor-tool" ], "patterns": [ { "kind": "Pattern", "slug": "agent-economics-moves-to-completed-work", "title": "Agent economics moves to completed work", "description": "The economically meaningful unit for agent systems is becoming cost per verified completed task rather than cost per token or model call.", "date": "2026-07-15", "topics": [ "agent-economics", "model-routing", "product-metrics" ], "source_count": 5, "metric": "high", "url": "https://newruntime.com/patterns/agent-economics-moves-to-completed-work/" } ], "field_notes": [ { "kind": "Field Note", "slug": "gemini-3-6-flash-agent-efficiency", "title": "Gemini 3.6 Flash Moves the Agent Race Toward Cost per Task", "description": "Google's Gemini 3.6 Flash release frames the model race around token efficiency, built-in computer use, and specialized cyber agents rather than raw chat intelligence alone.", "date": "2026-07-22", "topics": [ "gemini", "agent-economics", "coding-agents" ], "source_count": 1, "url": "https://newruntime.com/posts/gemini-3-6-flash-agent-efficiency/" } ], "raw_signals": [ { "kind": "Raw Signal", "slug": "fable-cost-per-completed-task", "title": "A pricier model can be cheaper per completed task", "description": "Cognition reports that Fable 5 completed coding work with fewer steps and output tokens than its previous lead model.", "date": "2026-07-15", "topics": [ "agent-economics", "coding-agents", "model-routing" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/fable-cost-per-completed-task/" }, { "kind": "Raw Signal", "slug": "claude-code-model-and-effort-are-separate-controls", "title": "Claude Code separates model choice from effort", "description": "Anthropic exposes model selection and effort level as different controls for capability, token use, latency, and persistence.", "date": "2026-07-13", "topics": [ "coding-agents", "model-routing", "agent-economics" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/claude-code-model-and-effort-are-separate-controls/" }, { "kind": "Raw Signal", "slug": "databricks-benchmarks-real-coding-agent-economics", "title": "Databricks benchmarks coding agents on its own codebase", "description": "Databricks evaluates agents on fresh internal pull-request tasks and measures success alongside runtime, tokens, and cost.", "date": "2026-07-12", "topics": [ "coding-agents", "evals", "agent-economics" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/databricks-benchmarks-real-coding-agent-economics/" }, { "kind": "Raw Signal", "slug": "advisor-model-guides-cheaper-executor", "title": "An advisor model can guide a cheaper executor", "description": "The advisor-tool pattern lets a fast executor request bounded analysis from a stronger model while keeping control of the task loop.", "date": "2026-07-09", "topics": [ "model-routing", "orchestration", "agent-economics" ], "source_count": 2, "metric": "notable", "url": "https://newruntime.com/signals/advisor-model-guides-cheaper-executor/" }, { "kind": "Raw Signal", "slug": "sonnet-moves-agent-capability-downmarket", "title": "Sonnet moves agent capability down the price curve", "description": "Anthropic positions Claude Sonnet 5 for planning, terminal work, browser use, and multi-step agent tasks at a lower tier.", "date": "2026-07-01", "topics": [ "models", "agent-economics", "coding-agents" ], "source_count": 1, "metric": "notable", "url": "https://newruntime.com/signals/sonnet-moves-agent-capability-downmarket/" }, { "kind": "Raw Signal", "slug": "ai-spend-measured-per-successful-task", "title": "AI spend should be measured per successful task", "description": "Ramp argues for allocating AI cost by use case, owner, completed outcome, failures, retries, review effort, and latency.", "date": "2026-06-27", "topics": [ "agent-economics", "ai-strategy", "product-metrics" ], "source_count": 1, "metric": "structural", "url": "https://newruntime.com/signals/ai-spend-measured-per-successful-task/" } ], "atlas_records": [] } ] }