[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-anthropic-models-build-better-agent-endurance-en":3,"article-related-anthropic-models-build-better-agent-endurance-en":30,"series-ai-agent-5115e111-8a16-41df-a00b-529be3f2c3c9":79},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":22,"views":26,"created_at":27,"published_at":28,"topic_cluster_id":29},"5115e111-8a16-41df-a00b-529be3f2c3c9","anthropic-models-build-better-agent-endurance-en","Anthropic Models: Build Better Agent Endurance","\u003Cp>When an \u003Ca href=\"\u002Ftag\u002Fagent\">agent\u003C\u002Fa> keeps pushing past the point where a human wants to step in, workflows can become slow, noisy, and hard to control. This guide shows how to shape prompts, stop conditions, and handoff rules so \u003Ca href=\"\u002Ftag\u002Fanthropic\">Anthropic\u003C\u002Fa>-based agents stay useful without running wild.\u003C\u002Fp>\u003Cp data-speakable=\"summary\">This guide shows how to control persistent Anthropic agents with clear stop rules.\u003C\u002Fp>\u003Ch2>Before you start\u003C\u002Fh2>\u003Cul>\u003Cli>An Anthropic account with API access\u003C\u002Fli>\u003Cli>An API key for Claude models\u003C\u002Fli>\u003Cli>Node.js 20+ or Python 3.11+\u003C\u002Fli>\u003Cli>A test project with at least one task that can be safely repeated\u003C\u002Fli>\u003Cli>Basic familiarity with prompt design and tool calling\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Step 1: Define the agent goal\u003C\u002Fh2>\u003Cp>Goal: turn one vague instruction into a task with a clear finish line, so the agent knows when to stop instead of improvising endlessly.\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1784102584376-pw9z.png\" alt=\"Anthropic Models: Build Better Agent Endurance\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>Write a single objective that includes scope, success criteria, and a hard stop. Keep it short enough to fit in one prompt block.\u003C\u002Fp>\u003Cpre>\u003Ccode>Goal: Draft a 5-bullet summary of the incident report, then stop and ask for approval before any rewrite.\u003C\u002Fcode>\u003C\u002Fpre>\u003Cp>You should see the agent complete the summary and then pause for confirmation, rather than continuing into edits or side tasks.\u003C\u002Fp>\u003Ch2>Step 2: Add stop conditions\u003C\u002Fh2>\u003Cp>Goal: make termination explicit, so the model has a rule for exiting even when it wants to keep trying alternative approaches.\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1784102579875-xxvc.png\" alt=\"Anthropic Models: Build Better Agent Endurance\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>Use a stop rule that names the trigger, the action, and the fallback. If the task needs human input, say so directly.\u003C\u002Fp>\u003Cpre>\u003Ccode>If you need missing information, stop after one attempt and ask the user for the exact field you need.\u003C\u002Fcode>\u003C\u002Fpre>\u003Cp>You should see the agent ask for help once instead of looping through repeated guesses or workaround attempts.\u003C\u002Fp>\u003Ch2>Step 3: Limit tool retries\u003C\u002Fh2>\u003Cp>Goal: prevent endless tool use by capping retries and defining what counts as a failed attempt.\u003C\u002Fp>\u003Cp>Set a maximum retry count for each tool and tell the agent what to do after the limit is reached. This works well for search, file edits, and \u003Ca href=\"\u002Ftag\u002Fapi\">API\u003C\u002Fa> calls.\u003C\u002Fp>\u003Cpre>\u003Ccode>Tool retry policy: 2 attempts max per tool. After that, return the last error and request human guidance.\u003C\u002Fcode>\u003C\u002Fpre>\u003Cp>You should see the agent stop after the second failed attempt and surface the error instead of reissuing the same call.\u003C\u002Fp>\u003Ch2>Step 4: Separate work from approval\u003C\u002Fh2>\u003Cp>Goal: keep execution and decision-making in different phases, so the agent does not treat every prompt as a command to finish everything at once.\u003C\u002Fp>\u003Cp>Structure the workflow as draft, review, and approve. The agent should only move to the next phase when the current phase is explicitly accepted.\u003C\u002Fp>\u003Cpre>\u003Ccode>Phase 1: propose options. Phase 2: wait for approval. Phase 3: execute only the approved option.\u003C\u002Fcode>\u003C\u002Fpre>\u003Cp>You should see the agent present options first and wait, rather than choosing a path and running ahead.\u003C\u002Fp>\u003Ch2>Step 5: Log failure patterns\u003C\u002Fh2>\u003Cp>Goal: capture where the agent overcommits, so you can tune prompts and policies based on real behavior instead of guesswork.\u003C\u002Fp>\u003Cp>Record the prompt, tool calls, retries, and the exact point where the agent should have stopped. Review these logs after each test run.\u003C\u002Fp>\u003Cpre>\u003Ccode>Log fields: task_id, prompt, tool_name, retry_count, stop_reason, human_override.\u003C\u002Fcode>\u003C\u002Fpre>\u003Cp>You should see repeat failure patterns, such as ignored stop cues or excessive retries, become easy to spot in the logs.\u003C\u002Fp>\u003Ch2>Common mistakes\u003C\u002Fh2>\u003Cul>\u003Cli>Writing goals that are too broad. Fix: add a measurable output and a clear stop point.\u003C\u002Fli>\u003Cli>Allowing unlimited retries. Fix: cap retries per tool and define a fallback response.\u003C\u002Fli>\u003Cli>Mixing approval with execution. Fix: split the workflow into draft, review, and approve phases.\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>What's next\u003C\u002Fh2>\u003Cp>Once your agent can stop cleanly, move on to evaluation harnesses, policy-based tool routing, and human-in-the-loop review flows so persistence becomes reliability instead of drift.\u003C\u002Fp>","A practical guide to designing agent prompts and stop rules around Anthropic’s persistent execution style.","www.zhihu.com","https:\u002F\u002Fwww.zhihu.com\u002Fquestion\u002F2059361484203996542",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1784102584376-pw9z.png","ai-agent","en","02bfe363-1ca6-4c20-be08-34d4ad532b71",[17,18,19,20,21],"Anthropic","Claude","prompt design","agent control","tool calling",[23,24,25],"Clear stop rules matter as much as task instructions for persistent agents.","Retry caps and phase separation reduce runaway tool use.","Logging failure points makes agent behavior easier to tune over time.",1,"2026-07-15T08:02:32.41246+00:00","2026-07-15T08:02:32.405+00:00","3ef18e30-4319-4b8c-aacb-cf5d9fba35de",{"tags":31,"relatedLang":38,"relatedPosts":42},[32,34,36],{"name":21,"slug":33},"tool-calling",{"name":17,"slug":35},"anthropic",{"name":18,"slug":37},"claude",{"id":15,"slug":39,"title":40,"language":41},"anthropic-model-task-persistence-tuning-zh","Anthropic 任務耐力調校指南","zh",[43,49,55,61,67,73],{"id":44,"slug":45,"title":46,"cover_image":47,"image_url":47,"created_at":48,"category":13},"6a959f48-0c84-4717-a858-d02c8bd26fce","googles-gemini-enterprise-agent-platform-cloud-service-en","Google’s Gemini Enterprise Agent Platform makes agents a cloud servic…","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1784059365473-8gbb.png","2026-07-14T20:02:25.392208+00:00",{"id":50,"slug":51,"title":52,"cover_image":53,"image_url":53,"created_at":54,"category":13},"3f5dcc12-bb78-4102-817b-f08f9adbe974","workbuddy-harness-engineering-matters-more-than-model-size-en","WorkBuddy Proves Harness Engineering Matters More Than Model Size","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783987381886-gvmo.png","2026-07-14T00:02:31.980136+00:00",{"id":56,"slug":57,"title":58,"cover_image":59,"image_url":59,"created_at":60,"category":13},"9c82943c-4f7b-4ada-8498-ce42b951a381","perplexity-teammate-coding-agent-strategy-en","Perplexity should build Teammate as a coding agent, not a copilot","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783589576551-x3my.png","2026-07-09T09:32:26.11108+00:00",{"id":62,"slug":63,"title":64,"cover_image":65,"image_url":65,"created_at":66,"category":13},"41ee43a5-607f-4990-9c24-90cc9a3d3366","hp-adopts-openai-frontier-global-operations-en","HP adopts OpenAI Frontier across global operations","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783362788376-y0kc.png","2026-07-06T18:32:31.325098+00:00",{"id":68,"slug":69,"title":70,"cover_image":71,"image_url":71,"created_at":72,"category":13},"ef1e437c-081d-4a40-bfdb-6370936f9442","build-production-vector-db-rag-pipeline-en","Build a production vector DB for RAG","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783258381626-nidw.png","2026-07-05T13:32:23.908316+00:00",{"id":74,"slug":75,"title":76,"cover_image":77,"image_url":77,"created_at":78,"category":13},"14f0f33c-469e-4996-be95-611f96cbffeb","ornith-1-agent-coding-server-template-en","Ornith-1 turns agent coding into a server","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783085625513-8n0k.png","2026-07-03T13:33:20.503272+00:00",[80,85,90,95,100,105,110,115,120,125],{"id":81,"slug":82,"title":83,"created_at":84},"03db8de8-8dc2-4ac1-9cf7-898782efbb1f","anthropic-claude-ai-agent-task-automation-en","Anthropic's Claude AI Agent: A New Era of Task Automation","2026-03-25T16:25:06.513026+00:00",{"id":86,"slug":87,"title":88,"created_at":89},"045d1abc-190d-4594-8c95-91e2a26f0c5a","googles-2026-ai-agent-report-decoded-en","Google’s 2026 AI Agent Report, Decoded","2026-03-26T11:15:23.046616+00:00",{"id":91,"slug":92,"title":93,"created_at":94},"e64aba21-254b-4f93-aa21-837484bb52ec","kimi-k25-review-stronger-still-not-legend-en","Kimi K2.5 review: stronger, still not a legend","2026-03-27T07:15:55.385951+00:00",{"id":96,"slug":97,"title":98,"created_at":99},"30dfb781-a1b2-4add-aebe-b3df40247c37","claude-code-controls-mac-desktop-en","Claude Code now controls your Mac desktop","2026-03-28T03:01:59.384091+00:00",{"id":101,"slug":102,"title":103,"created_at":104},"254405b6-7833-4800-8e13-f5196deefbe6","cloudflare-100x-faster-ai-agent-sandbox-en","Cloudflare’s 100x Faster AI Agent Sandbox","2026-03-28T03:09:44.356437+00:00",{"id":106,"slug":107,"title":108,"created_at":109},"04f29b7f-9b91-4306-89a7-97d725e6e1ba","openai-backs-isara-agent-swarm-bet-en","OpenAI backs Isara’s agent-swarm bet","2026-03-28T03:15:27.849766+00:00",{"id":111,"slug":112,"title":113,"created_at":114},"3b0bf479-e4ae-4703-9666-721a7e0cdb91","openai-plan-automated-ai-researcher-en","OpenAI’s plan for an automated AI researcher","2026-03-28T03:17:42.312819+00:00",{"id":116,"slug":117,"title":118,"created_at":119},"fe91bce0-b85d-4efa-a207-24ae9939c29f","harness-engineering-ai-agent-reliability-2026","Harness Engineering: From Bridle to Operating System, The Missing Link in AI Agent Reliability","2026-03-31T06:36:55.648751+00:00",{"id":121,"slug":122,"title":123,"created_at":124},"7a09007d-820f-43b3-8607-8ad1bfcb94c8","mcp-explained-from-prompts-to-production-en","MCP Explained: From Prompts to Production","2026-04-01T09:24:40.089177+00:00",{"id":126,"slug":127,"title":128,"created_at":129},"116d5ee9-a4f1-4b5a-aac5-5d035dd22bbe","amazon-bedrock-agents-multi-agent-workflows-en","Amazon Bedrock Agents Gets Multi-Agent Workflows","2026-04-01T09:30:30.197685+00:00"]