[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7B3PRRG11A17DT19J3RP97N":3,"skill-same:01M7B3PRRG11A17DT19J3RP97N":46},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":29,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M7B3PRRG11A17DT19J3RP97N","agent-builder",null,"Use when building a new agent harness, designing tool systems, or structuring multi-agent workflows. Provides patterns, templates, and decision trees for harness engineering.","javakishore-veleti\u002Fclaude-code-enterprise-it-harnessing","javakishore-veleti","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F6062900?v=4","01M7B3EG9GDQBN83N7ZWJ7JGAX","Kishore Veleti","01M7B3EGA3ZH43Y1XHGX5N50BK",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing\u002Fblob\u002Fmain\u002Fskills\u002Fagent-builder\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: agent-builder\ndescription: Use when building a new agent harness, designing tool systems, or structuring multi-agent workflows. Provides patterns, templates, and decision trees for harness engineering.\n---\n\n# Agent Builder Skill\n\n## When to use this skill\nLoad this skill when the user wants to:\n- Build a new agent from scratch\n- Design a tool for an agent\n- Structure a multi-agent system\n- Debug an agent loop that isn't working\n- Choose between agent architectures\n\n## Core principle\n\nThe agent is always the model. Your job is the harness.\n\n```\nHarness = Tools + Knowledge + Observation + Action + Permissions\n```\n\nNever try to encode intelligence in your harness code. Give the model\nclean tools, clear context, and get out of the way.\n\n## The minimal agent (always start here)\n\n```python\nfrom anthropic import Anthropic\nclient = Anthropic()\n\ndef agent_loop(messages, tools, dispatch, system):\n    while True:\n        response = client.messages.create(\n            model=\"claude-sonnet-4-20250514\",\n            system=system, messages=messages,\n            tools=tools, max_tokens=8000,\n        )\n        messages.append({\"role\": \"assistant\", \"content\": response.content})\n        if response.stop_reason != \"tool_use\":\n            return\n        results = []\n        for block in response.content:\n            if block.type == \"tool_use\":\n                output = dispatch[block.name](block.input)\n                results.append({\"type\": \"tool_result\",\n                                 \"tool_use_id\": block.id, \"content\": output})\n        messages.append({\"role\": \"user\", \"content\": results})\n```\n\nDo not add anything until you need it. Every mechanism should earn its place.\n\n## Tool design checklist\n\nBefore writing a tool, ask:\n- [ ] Is the name a verb? (bash, read, write — not \"file_manager\")\n- [ ] Does the description say WHEN to use it, not just what it does?\n- [ ] Is the input schema minimal? No optional fields unless truly needed.\n- [ ] Does it return plain text the model can reason about?\n- [ ] Does it have a hard timeout?\n- [ ] Is output truncated to a safe length (≤50k chars)?\n\n## Tool description formula\n\n```\n\"[Action verb] [what it does]. Use when [specific situation].\n[What it returns]. [Any important limits].\"\n```\n\nExample:\n```\n\"Read a file and return numbered lines. Use when you need to inspect\nfile content or reference specific line numbers. Returns up to 50,000\ncharacters. Use start_line\u002Fend_line for large files.\"\n```\n\n## Architecture decision tree\n\n```\nOne task, one user, no persistence needed?\n    → s01: minimal loop + bash\n\nNeed file read\u002Fwrite\u002Fsearch?\n    → s02: extended tool dispatch\n\nNeed the agent to plan before acting?\n    → s03: add todo_write tool\n\nTask too big for one context window?\n    → s04: subagent isolation\n\nNeed domain-specific knowledge?\n    → s05: skill loading\n\nLong-running session, context will overflow?\n    → s06: compression + memory file\n\nComplex multi-step project spanning sessions?\n    → s07: task graph with dependencies\n\nSlow operations (builds, tests)?\n    → s08: background tasks\n\nWork that parallelises across specialties?\n    → s09+: agent teams with mailboxes\n\nNeed isolation between parallel tasks?\n    → s12\u002Fs23: git worktrees\n```\n\n## Common mistakes\n\n**Putting logic in the harness instead of trusting the model**\nBad:  `if \"error\" in output: retry_with_different_approach()`\nGood: return the error to the model and let it decide\n\n**Giant system prompts**\nBad:  5,000-word system prompt covering every scenario\nGood: load domain knowledge on-demand via skills (s05)\n\n**Blocking the loop on slow operations**\nBad:  `output = subprocess.run(\"npm test\", timeout=300)`\nGood: run in background thread, notify when done (s08)\n\n**Shared mutable state between subagents**\nBad:  subagents writing to the same dict\u002Ffile without locks\nGood: each subagent has its own isolated context (s04, s12)\n\n## Subagent pattern template\n\n```python\ndef spawn_subagent(prompt: str, tools=EXTENDED_TOOLS, dispatch=EXTENDED_DISPATCH) -> str:\n    messages = [{\"role\": \"user\", \"content\": prompt}]\n    while True:\n        response = client.messages.create(\n            model=MODEL, system=SUBAGENT_SYSTEM,\n            messages=messages, tools=tools, max_tokens=8000,\n        )\n        messages.append({\"role\": \"assistant\", \"content\": response.content})\n        if response.stop_reason != \"tool_use\":\n            break\n        results = dispatch_tools(response.content, dispatch)\n        messages.append({\"role\": \"user\", \"content\": results})\n    return \"\".join(b.text for b in messages[-1][\"content\"] if hasattr(b, \"text\"))\n```","# Agent Builder Skill\n\n## When to use this skill\nLoad this skill when the user wants to:\n- Build a new agent from scratch\n- Design a tool for an agent\n- Structure a multi-agent system\n- Debug an agent loop that isn't working\n- Choose between agent architectures\n\n## Core principle\n\nThe agent is always the model. Your job is the harness.\n\n```\nHarness = Tools + Knowledge + Observation + Action + Permissions\n```\n\nNever try to encode intelligence in your harness code. Give the model\nclean tools, clear context, and get out of the way.\n\n## The minimal agent (always start here)\n\n```python\nfrom anthropic import Anthropic\nclient = Anthropic()\n\ndef agent_loop(messages, tools, dispatch, system):\n    while True:\n        response = client.messages.create(\n            model=\"claude-sonnet-4-20250514\",\n            system=system, messages=messages,\n            tools=tools, max_tokens=8000,\n        )\n        messages.append({\"role\": \"assistant\", \"content\": response.content})\n        if response.stop_reason != \"tool_use\":\n            return\n        results = []\n        for block in response.content:\n            if block.type == \"tool_use\":\n                output = dispatch[block.name](block.input)\n                results.append({\"type\": \"tool_result\",\n                                 \"tool_use_id\": block.id, \"content\": output})\n        messages.append({\"role\": \"user\", \"content\": results})\n```\n\nDo not add anything until you need it. Every mechanism should earn its place.\n\n## Tool design checklist\n\nBefore writing a tool, ask:\n- [ ] Is the name a verb? (bash, read, write — not \"file_manager\")\n- [ ] Does the description say WHEN to use it, not just what it does?\n- [ ] Is the input schema minimal? No optional fields unless truly needed.\n- [ ] Does it return plain text the model can reason about?\n- [ ] Does it have a hard timeout?\n- [ ] Is output truncated to a safe length (≤50k chars)?\n\n## Tool description formula\n\n```\n\"[Action verb] [what it does]. Use when [specific situation].\n[What it returns]. [Any important limits].\"\n```\n\nExample:\n```\n\"Read a file and return numbered lines. Use when you need to inspect\nfile content or reference specific line numbers. Returns up to 50,000\ncharacters. Use start_line\u002Fend_line for large files.\"\n```\n\n## Architecture decision tree\n\n```\nOne task, one user, no persistence needed?\n    → s01: minimal loop + bash\n\nNeed file read\u002Fwrite\u002Fsearch?\n    → s02: extended tool dispatch\n\nNeed the agent to plan before acting?\n    → s03: add todo_write tool\n\nTask too big for one context window?\n    → s04: subagent isolation\n\nNeed domain-specific knowledge?\n    → s05: skill loading\n\nLong-running session, context will overflow?\n    → s06: compression + memory file\n\nComplex multi-step project spanning sessions?\n    → s07: task graph with dependencies\n\nSlow operations (builds, tests)?\n    → s08: background tasks\n\nWork that parallelises across specialties?\n    → s09+: agent teams with mailboxes\n\nNeed isolation between parallel tasks?\n    → s12\u002Fs23: git worktrees\n```\n\n## Common mistakes\n\n**Putting logic in the harness instead of trusting the model**\nBad:  `if \"error\" in output: retry_with_different_approach()`\nGood: return the error to the model and let it decide\n\n**Giant system prompts**\nBad:  5,000-word system prompt covering every scenario\nGood: load domain knowledge on-demand via skills (s05)\n\n**Blocking the loop on slow operations**\nBad:  `output = subprocess.run(\"npm test\", timeout=300)`\nGood: run in background thread, notify when done (s08)\n\n**Shared mutable state between subagents**\nBad:  subagents writing to the same dict\u002Ffile without locks\nGood: each subagent has its own isolated context (s04, s12)\n\n## Subagent pattern template\n\n```python\ndef spawn_subagent(prompt: str, tools=EXTENDED_TOOLS, dispatch=EXTENDED_DISPATCH) -> str:\n    messages = [{\"role\": \"user\", \"content\": prompt}]\n    while True:\n        response = client.messages.create(\n            model=MODEL, system=SUBAGENT_SYSTEM,\n            messages=messages, tools=tools, max_tokens=8000,\n        )\n        messages.append({\"role\": \"assistant\", \"content\": response.content})\n        if response.stop_reason != \"tool_use\":\n            break\n        results = dispatch_tools(response.content, dispatch)\n        messages.append({\"role\": \"user\", \"content\": results})\n    return \"\".join(b.text for b in messages[-1][\"content\"] if hasattr(b, \"text\"))\n```",{"name":5,"description":7},"main","d3dac4e9ef1de1599cf281253d693f122d426b78","skills\u002Fagent-builder","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti","https:\u002F\u002Fwww.youtube.com\u002Fchannel\u002FUCok0L9Tbjl7nzfwaEOXuBHQ","https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fvavkkishore","2026-08-30 05:44:55",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add javakishore-veleti\u002Fclaude-code-enterprise-it-harnessing --skill agent-builder",{"tool":38,"command":39},"bunx","bunx skills add javakishore-veleti\u002Fclaude-code-enterprise-it-harnessing --skill agent-builder",{"tool":41,"command":42},"pnpm","pnpm dlx skills add javakishore-veleti\u002Fclaude-code-enterprise-it-harnessing --skill agent-builder","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing\u002Fblob\u002Fmain\u002Fskills\u002Fagent-builder\u002FSKILL.md",true,[],[47,54,61,68,75,82],{"skillNo":48,"name":49,"nameZh":6,"description":50,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":51,"categories":52,"keywords":53,"publishedAt":19},"01M7B3PYZ0YFXF88H2VTF8NZR6","pdf","Use when working with PDF files — reading, extracting text or tables, merging, splitting, or filling forms. Provides correct library choices and common patterns.","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing\u002Fblob\u002Fmain\u002Fskills\u002Fpdf\u002FSKILL.md",[],[],{"skillNo":55,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M7B3PWD52QWXMG2AZKSEMHF4","code-review","Use when asked to review code, audit a file for bugs, check code quality, or suggest improvements. Provides a structured review process and issue categorisation.","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing\u002Fblob\u002Fmain\u002Fskills\u002Fcode-review\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":19},"01M7B3PP85ECYV7QN9XZ59XGDZ","ticket-sla-breach","Support ticket SLA breach","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing\u002Fblob\u002Fmain\u002Fenterprise_it_harnessing\u002Fsre_harnessing\u002Fskills\u002Fticket-sla-breach\u002FSKILL.md",[],[],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M7B3PKNKVJDYG17E2ZFQW6QN","quote-to-order-break","B2B quote-to-order break","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing\u002Fblob\u002Fmain\u002Fenterprise_it_harnessing\u002Fsre_harnessing\u002Fskills\u002Fquote-to-order-break\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M7B3PH7F2AJXP7FKHDWM07Q7","payments-decline-storm","Payments decline storm","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing\u002Fblob\u002Fmain\u002Fenterprise_it_harnessing\u002Fsre_harnessing\u002Fskills\u002Fpayments-decline-storm\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M7B3PEQ48V382CHNCXKXKW79","page-forex-oncall","Page FOREX markets on-call","https:\u002F\u002Fgithub.com\u002Fjavakishore-veleti\u002Fclaude-code-enterprise-it-harnessing\u002Fblob\u002Fmain\u002Fenterprise_it_harnessing\u002Fsre_harnessing\u002Fskills\u002Fpage-forex-oncall\u002FSKILL.md",[],[]]