{"id":"pg_vhb7nnooqypu","property_id":"prp_buyvj9b2can9","domain_id":"dom_aplqvkihzdue","root_domain":"anthropic.com","locale":"en","url":"https://anthropic.com/engineering","url_hash":"d40bbcf868392b0e920c15e40ed72c8f47073f2c49aac66752a1180438c9daf4","path":"/engineering","depth":1,"canonical_url":"https://anthropic.com/engineering","canonical_page_id":null,"page_type":"landing","page_type_guess":"landing","state":"indexed","skip_reason":null,"priority":"0.9","discovered_via":"sitemap","first_seen_at":"2026-08-27T09:06:23.909Z","last_seen_at":"2026-08-27T09:07:05.847Z","http_status":200,"content_hash":"380bce76916f0164ff043f6b96b098fb4e96723ccf830df1e8b340582f685127","title":"Engineering","description":"Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.","author":null,"published_at":null,"modified_at":null,"word_count":557,"reading_time_sec":152,"lang_detected":"en","images":[],"links_out":null,"markdown_source":"readability","refresh_interval":"monthly","last_fetched_at":"2026-08-27T09:06:46.906Z","next_refresh_at":"2026-09-26T09:06:46.906Z","runs_count":1,"changes_count":0,"last_changed_at":null,"fail_count":0,"last_error":null,"created_at":"2026-08-27T09:06:23.909Z","updated_at":"2026-08-27T09:06:23.909Z","inlinks":14,"inlinks_body":1,"outlinks":58,"outlinks_external":5,"anchor_terms":["Engineering","Engineering at Anthropic"],"template_score":"0.93333333333333333333","cash":"0.0227006466515116999","cash_total":"0.3618109087523618219","pagerank":null,"is_orphan":false,"link_confirmed":true,"discovered_wave":0,"embedded_hash":"380bce76916f0164ff043f6b96b098fb4e96723ccf830df1e8b340582f685127","embedded_at":"2026-08-27T09:13:51.706Z","domain_name":"Anthropic","brand_color":null,"domain_rank":623,"spam_score":11,"markdown":"![How we contain Claude across products](https://www-cdn.anthropic.com/images/4zrzovbb/website/47d14a71a7a759af39e1bc36ee68d65eb16ad74d-1000x1000.svg)\n\n[\n\n![How we contain Claude across products](https://www-cdn.anthropic.com/images/4zrzovbb/website/47d14a71a7a759af39e1bc36ee68d65eb16ad74d-1000x1000.svg)\n\nFeatured\n\n## How we contain Claude across products\n\nAs agents grow more capable, so does their potential blast radius. The engineering question is how to cap it. Here’s what we’ve learned building containment for claude.ai, Claude Code, and Cowork.\n\n](https://anthropic.com/engineering/how-we-contain-claude)\n\n[\n\n![An update on recent Claude Code quality reports](https://www-cdn.anthropic.com/images/4zrzovbb/website/259cb9466c31eee1bd312c230c1ab95c844da488-500x500.svg)\n\n### An update on recent Claude Code quality reports\n\nApr 23, 2026\n\n](https://anthropic.com/engineering/april-23-postmortem)\n\n[\n\n![Scaling Managed Agents: Decoupling the brain from the hands](https://www-cdn.anthropic.com/images/4zrzovbb/website/7675e9c4ed4c7a8fe2e4df296fd1c4adac5b652b-1200x1200.svg)\n\n### Scaling Managed Agents: Decoupling the brain from the hands\n\nApr 08, 2026\n\n](https://anthropic.com/engineering/managed-agents)\n\n[\n\n![How we built Claude Code auto mode: a safer way to skip permissions](https://www-cdn.anthropic.com/images/4zrzovbb/website/b87185e4d533134bc3f9b949a874396dcfcb2e80-500x500.svg)\n\n### How we built Claude Code auto mode: a safer way to skip permissions\n\nMar 25, 2026\n\n](https://anthropic.com/engineering/claude-code-auto-mode)\n\n[\n\n![Harness design for long-running application development](https://www-cdn.anthropic.com/images/4zrzovbb/website/af0acebfbd57ac4b26ae7d7ae124d7326a3e47e4-1200x1200.svg)\n\n### Harness design for long-running application development\n\nMar 24, 2026\n\n](https://anthropic.com/engineering/harness-design-long-running-apps)\n\n[\n\n![Eval awareness in Claude Opus 4.6’s BrowseComp performance](https://www-cdn.anthropic.com/images/4zrzovbb/website/641d32b3291956d595c7e820d5bf94c5f44baa28-500x500.svg)\n\n### Eval awareness in Claude Opus 4.6’s BrowseComp performance\n\nMar 06, 2026\n\n](https://anthropic.com/engineering/eval-awareness-browsecomp)\n\n[\n\n### Quantifying infrastructure noise in agentic coding evals\n\nFeb 05, 2026\n\n](https://anthropic.com/engineering/infrastructure-noise)\n\n[\n\n![Building a C compiler with a team of parallel Claudes](https://www-cdn.anthropic.com/images/4zrzovbb/website/44e93e074d53285f64ff717365b04c4a2164a445-1200x1200.svg)\n\n### Building a C compiler with a team of parallel Claudes\n\nFeb 05, 2026\n\n](https://anthropic.com/engineering/building-c-compiler)\n\n[\n\n![Designing AI-resistant technical evaluations](https://www-cdn.anthropic.com/images/4zrzovbb/website/dc34c3eeae881b105ef652d5630d84de6a1fa01a-1200x1200.svg)\n\n### Designing AI-resistant technical evaluations\n\nJan 21, 2026\n\n](https://anthropic.com/engineering/AI-resistant-technical-evaluations)\n\n[\n\n![Demystifying evals for AI agents](https://www-cdn.anthropic.com/images/4zrzovbb/website/b87185e4d533134bc3f9b949a874396dcfcb2e80-500x500.svg)\n\n### Demystifying evals for AI agents\n\nJan 09, 2026\n\n](https://anthropic.com/engineering/demystifying-evals-for-ai-agents)\n\n[\n\n![Effective harnesses for long-running agents](https://www-cdn.anthropic.com/images/4zrzovbb/website/c041b5e0498972014414a7c3d044727982f26bde-500x500.svg)\n\n### Effective harnesses for long-running agents\n\nNov 26, 2025\n\n](https://anthropic.com/engineering/effective-harnesses-for-long-running-agents)\n\n[\n\n![Illustration for advanced tool use article.](https://www-cdn.anthropic.com/images/4zrzovbb/website/2aa849e93e76ae567502dcae2db8921062531fa1-500x500.svg)\n\n### Introducing advanced tool use on the Claude Developer Platform\n\nNov 24, 2025\n\n](https://anthropic.com/engineering/advanced-tool-use)\n\n[\n\n![Code execution with MCP: Building more efficient agents](https://www-cdn.anthropic.com/images/4zrzovbb/website/848e961961a97ada3a7edb2d1d17378792c3288d-500x500.svg)\n\n### Code execution with MCP: Building more efficient agents\n\nNov 04, 2025\n\n](https://anthropic.com/engineering/code-execution-with-mcp)\n\n[\n\n![Beyond permission prompts: making Claude Code more secure and autonomous](https://www-cdn.anthropic.com/images/4zrzovbb/website/33d37e1ae729f4e960d11fecf143ac14c0fb369d-500x500.svg)\n\n### Beyond permission prompts: making Claude Code more secure and autonomous\n\nOct 20, 2025\n\n](https://anthropic.com/engineering/claude-code-sandboxing)\n\n[\n\n![Equipping agents for the real world with Agent Skills](https://www-cdn.anthropic.com/images/4zrzovbb/website/b4fe0845239779c6fc1e045edb6272c3f500944a-500x500.svg)\n\n### Equipping agents for the real world with Agent Skills\n\nOct 16, 2025\n\n](https://anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills)\n\n[\n\n![Effective context engineering for AI agents](https://www-cdn.anthropic.com/images/4zrzovbb/website/a048404a96b599af98c05da5bdd1db07222e4e7b-500x500.svg)\n\n### Effective context engineering for AI agents\n\nSep 29, 2025\n\n](https://anthropic.com/engineering/effective-context-engineering-for-ai-agents)\n\n[\n\n![A postmortem of three recent issues](https://www-cdn.anthropic.com/images/4zrzovbb/website/33d37e1ae729f4e960d11fecf143ac14c0fb369d-500x500.svg)\n\n### A postmortem of three recent issues\n\nSep 17, 2025\n\n](https://anthropic.com/engineering/a-postmortem-of-three-recent-issues)\n\n[\n\n![This is an abstract illustration for the Eng Blog article, Writing effective tools for agents -- with agents.](https://www-cdn.anthropic.com/images/4zrzovbb/website/b4fe0845239779c6fc1e045edb6272c3f500944a-500x500.svg)\n\n### Writing effective tools for agents — with agents\n\nSep 11, 2025\n\n](https://anthropic.com/engineering/writing-tools-for-agents)\n\n[\n\n![Desktop Extensions: One-click MCP server installation for Claude Desktop](https://www-cdn.anthropic.com/images/4zrzovbb/website/dde35f184e14e5c37b0b3ab5a1c0bbad06ac123b-500x500.svg)\n\n### Desktop Extensions: One-click MCP server installation for Claude Desktop\n\nJun 26, 2025\n\n](https://anthropic.com/engineering/desktop-extensions)\n\n[\n\n![How we built our multi-agent research system](https://www-cdn.anthropic.com/images/4zrzovbb/website/848e961961a97ada3a7edb2d1d17378792c3288d-500x500.svg)\n\n### How we built our multi-agent research system\n\nJun 13, 2025\n\n](https://anthropic.com/engineering/multi-agent-research-system)\n\n[\n\n![Claude Code: Best practices for agentic coding](https://www-cdn.anthropic.com/images/4zrzovbb/website/c423cdaa6733c03a5d10f38c76e1ecf1900c6716-1200x1200.svg)\n\n### Claude Code: Best practices for agentic coding\n\nApr 18, 2025\n\n](https://anthropic.com/engineering/claude-code-best-practices)\n\n[\n\n![Abstract shapes illustrating Anthropic's Engineering Blog](https://www-cdn.anthropic.com/images/4zrzovbb/website/461ea9ed02230ba02ab830e5a5b23df66ea23bc8-1200x1200.svg)\n\n### The \"think\" tool: Enabling Claude to stop and think in complex tool use situations\n\nMar 20, 2025\n\n](https://anthropic.com/engineering/claude-think-tool)\n\n[\n\n![Raising the bar on SWE-bench Verified with Claude 3.5 Sonnet](https://www-cdn.anthropic.com/images/4zrzovbb/website/ef693b8c4ebfcead4e17af7bd87b66f8bc70b8cc-1200x1200.svg)\n\n### Raising the bar on SWE-bench Verified with Claude 3.5 Sonnet\n\nJan 06, 2025\n\n](https://anthropic.com/engineering/swe-bench-sonnet)\n\n[\n\n![Building effective agents](https://www-cdn.anthropic.com/images/4zrzovbb/website/14b20fce6e93c79be47352da0fa4bebd597ebfa8-1200x1200.svg)\n\n### Building effective agents\n\nDec 19, 2024\n\n](https://anthropic.com/engineering/building-effective-agents)\n\n[\n\n![Introducing Contextual Retrieval](https://www-cdn.anthropic.com/images/4zrzovbb/website/44e93e074d53285f64ff717365b04c4a2164a445-1200x1200.svg)\n\n### Introducing Contextual Retrieval\n\nSep 19, 2024\n\n](https://anthropic.com/engineering/contextual-retrieval)","storage":{"htmlKey":"stealpage/html/anthropic.com/pg_vhb7nnooqypu/run_6mu7dgssvy6f.html.gz","markdownKey":"stealpage/md/anthropic.com/pg_vhb7nnooqypu/run_6mu7dgssvy6f.md.gz"},"facets":{"page_id":"pg_vhb7nnooqypu","domain_id":"dom_aplqvkihzdue","root_domain":"anthropic.com","locale":"en","format":"research","depth":1,"technicality":2,"promotional":4,"originality":4,"audience":"general","funnel_stage":"awareness","intent":"informational","evidence":"none","tone":"academic","has_code":false,"code_langs":[],"has_data":true,"has_tables":false,"has_images":false,"has_video":false,"has_faq":false,"external_links":50,"heading_count":25,"topics":["ai safety","machine learning","software engineering"],"keywords":["containment"," Claude","agentic coding","evals","managed agents"],"entities":{},"takeaways":["The page serves as a hub for Anthropic's engineering blog posts, focusing on technical aspects of AI development and safety.","It highlights the company's work on containment strategies for AI agents and improving their reliability and security.","The content includes updates on specific projects like Claude Code, managed agents, and evals for AI systems.","The page demonstrates Anthropic's commitment to transparency and sharing technical insights about their AI research.","It provides a comprehensive overview of the engineering challenges and solutions related to building safe and effective AI systems."],"summary":"The page provides a collection of technical engineering posts from Anthropic about their work on AI safety, agent systems, and model evaluation.","ai_likelihood":"0","shelf_life":"dated","quality_score":"0.4699","fingerprint":"380bce76916f0164ff043f6b96b098fb4e96723ccf830df1e8b340582f685127","model":"qwen/qwen3-30b-a3b","classified_at":"2026-08-27T09:11:21.613Z","created_at":"2026-08-27T09:11:21.613Z","updated_at":"2026-08-27T09:11:21.613Z"},"assets":[{"kind":"image","src_url":"https://cdn.sanity.io/images/4zrzovbb/website/6d4a0d28992ade92d6fa63646fd9c9d318245c6c-2400x1260.jpg","r2_key":null,"alt":null,"caption":null,"width":null,"height":null,"is_hero":true,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/47d14a71a7a759af39e1bc36ee68d65eb16ad74d-1000x1000.svg","r2_key":null,"alt":"How we contain Claude across products","caption":null,"width":1000,"height":1000,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/259cb9466c31eee1bd312c230c1ab95c844da488-500x500.svg","r2_key":null,"alt":"An update on recent Claude Code quality reports","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/7675e9c4ed4c7a8fe2e4df296fd1c4adac5b652b-1200x1200.svg","r2_key":null,"alt":"Scaling Managed Agents: Decoupling the brain from the hands","caption":null,"width":1200,"height":1200,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/b87185e4d533134bc3f9b949a874396dcfcb2e80-500x500.svg","r2_key":null,"alt":"How we built Claude Code auto mode: a safer way to skip permissions","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/af0acebfbd57ac4b26ae7d7ae124d7326a3e47e4-1200x1200.svg","r2_key":null,"alt":"Harness design for long-running application development","caption":null,"width":1200,"height":1200,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/641d32b3291956d595c7e820d5bf94c5f44baa28-500x500.svg","r2_key":null,"alt":"Eval awareness in Claude Opus 4.6’s BrowseComp performance","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/44e93e074d53285f64ff717365b04c4a2164a445-1200x1200.svg","r2_key":null,"alt":"Building a C compiler with a team of parallel Claudes","caption":null,"width":1200,"height":1200,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/dc34c3eeae881b105ef652d5630d84de6a1fa01a-1200x1200.svg","r2_key":null,"alt":"Designing AI-resistant technical evaluations","caption":null,"width":1200,"height":1200,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/c041b5e0498972014414a7c3d044727982f26bde-500x500.svg","r2_key":null,"alt":"Effective harnesses for long-running agents","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/2aa849e93e76ae567502dcae2db8921062531fa1-500x500.svg","r2_key":null,"alt":"Illustration for advanced tool use article.","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/848e961961a97ada3a7edb2d1d17378792c3288d-500x500.svg","r2_key":null,"alt":"Code execution with MCP: Building more efficient agents","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/33d37e1ae729f4e960d11fecf143ac14c0fb369d-500x500.svg","r2_key":null,"alt":"Beyond permission prompts: making Claude Code more secure and autonomous","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/b4fe0845239779c6fc1e045edb6272c3f500944a-500x500.svg","r2_key":null,"alt":"Equipping agents for the real world with Agent Skills","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/a048404a96b599af98c05da5bdd1db07222e4e7b-500x500.svg","r2_key":null,"alt":"Effective context engineering for AI agents","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/dde35f184e14e5c37b0b3ab5a1c0bbad06ac123b-500x500.svg","r2_key":null,"alt":"Desktop Extensions: One-click MCP server installation for Claude Desktop","caption":null,"width":500,"height":500,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/c423cdaa6733c03a5d10f38c76e1ecf1900c6716-1200x1200.svg","r2_key":null,"alt":"Claude Code: Best practices for agentic coding","caption":null,"width":1200,"height":1200,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/461ea9ed02230ba02ab830e5a5b23df66ea23bc8-1200x1200.svg","r2_key":null,"alt":"Abstract shapes illustrating Anthropic's Engineering Blog","caption":null,"width":1200,"height":1200,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/ef693b8c4ebfcead4e17af7bd87b66f8bc70b8cc-1200x1200.svg","r2_key":null,"alt":"Raising the bar on SWE-bench Verified with Claude 3.5 Sonnet","caption":null,"width":1200,"height":1200,"is_hero":false,"zone":"main"},{"kind":"image","src_url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/14b20fce6e93c79be47352da0fa4bebd597ebfa8-1200x1200.svg","r2_key":null,"alt":"Building effective agents","caption":null,"width":1200,"height":1200,"is_hero":false,"zone":"main"}],"shots":[],"history":[{"id":"run_6mu7dgssvy6f","started_at":"2026-08-27T09:06:46.834Z","http_status":200,"fetcher":"direct","changed":false,"diff":null,"word_count":557}],"citedBy":[{"id":"pg_zobp_qqgrfys","url":"https://anthropic.com/","title":"Home \\ Anthropic","anchors":["Engineering","Engineering at Anthropic"],"zone":"main"}]}