[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"workflow-ms-agent-lightweight-agent-framework-with-mcp-376e2fa5":3,"seo:featured-workflow:376e2fa5-a3f0-52fc-a8df-f212825e14a4:en":38,"workflow-related-ms-agent-lightweight-agent-framework-with-mcp-376e2fa5-376e2fa5-a3f0-52fc-a8df-f212825e14a4":82},{"id":4,"uuid":5,"slug":6,"title":7,"description":8,"author_id":9,"author_name":10,"author_avatar":11,"token_estimate":12,"time_saved":12,"model_used":13,"fork_count":12,"vote_count":12,"view_count":14,"parent_id":12,"parent_uuid":13,"lang_type":15,"steps":16,"tags":23,"has_voted":29,"visibility":19,"share_token":13,"is_featured":12,"content_hash":30,"asset_kind":27,"target_tools":31,"install_mode":35,"entrypoint":36,"risk_profile":37,"dependencies":39,"verification":45,"agent_metadata":48,"agent_fit":59,"trust":70,"provenance":78,"created_at":80,"updated_at":81},3309,"376e2fa5-a3f0-52fc-a8df-f212825e14a4","ms-agent-lightweight-agent-framework-with-mcp","MS-Agent — Lightweight Agent Framework with MCP","MS-Agent is a lightweight agent framework with MCP configs; install via pip and start with an LLMAgent example wired to Streamable HTTP MCP endpoints.","8a910fec-3180-11f1-9bc6-00163e2b0d79","Agent Toolkit","https:\u002F\u002Ftokrepo.com\u002Fapple-touch-icon.png",0,"",15,"en",[17],{"id":18,"step_order":19,"title":20,"description":13,"prompt_template":21,"variables":13,"depends_on":22,"expected_output":13},3872,1,"Asset","## Quick Use\n\n1. Install:\n   ```bash\n   pip install ms-agent\n   # or: pip install 'ms-agent[research]'\n   ```\n2. Set your ModelScope key:\n   ```bash\n   export MODELSCOPE_API_KEY={your_modelscope_api_key}\n   ```\n3. Run the MCP-wired example from the README (LLMAgent + `mcpServers`).\n\n## Intro\n\nMS-Agent is a lightweight agent framework with MCP configs; install via pip and start with an LLMAgent example wired to Streamable HTTP MCP endpoints.\n\n- **Best for:** teams who want an agent runner that can consume MCP configs without building a custom harness\n- **Works with:** Python; ModelScope API (per README); Streamable HTTP MCP endpoints\n- **Setup time:** 15–35 minutes\n\n## Practical Notes\n\n- Quant: install is `pip install ms-agent` with an optional `[research]` extra per README.\n- Quant: validate MCP wiring by calling one tool via an MCP endpoint and logging the round-trip latency (baseline for regressions).\n\n## How to use it as a harness (not a demo)\n\nThe README example is a good start, but production usage needs guardrails:\n\n1. **Pin MCP server configs** per environment (dev\u002Fstaging\u002Fprod) and store them in version control.\n2. **Separate credentials** from configs: env vars only, no secrets checked into repos.\n3. **Replayability**: keep a small set of “golden tasks” you can re-run after upgrades to detect regressions.\n\nIf your goal is evaluation\u002Fsecurity, wrap runs with:\n\n- input\u002Foutput logging,\n- tool-call allow-lists,\n- and a cost\u002Flatency budget per task.\n\n### FAQ\n\n**Q: Does it support MCP?**\nA: Yes. The README includes an example `mcpServers` config wired into LLMAgent.\n\n**Q: How do I keep runs reproducible?**\nA: Pin configs + keep golden tasks + log tool calls and outputs.\n\n**Q: What should I do first?**\nA: Wire one simple MCP server endpoint and run a single task end-to-end before adding more tools.\n\n## Source & Thanks\n\n> Source: https:\u002F\u002Fgithub.com\u002Fmodelscope\u002Fms-agent\n> License: Apache-2.0\n> GitHub stars: 4,244 · forks: 496\n\n---\n\n\u003C!-- ZH -->\n\n## 快速使用\n\n1. 安装：\n   ```bash\n   pip install ms-agent\n   # 或：pip install 'ms-agent[research]'\n   ```\n2. 配置 ModelScope Key：\n   ```bash\n   export MODELSCOPE_API_KEY={your_modelscope_api_key}\n   ```\n3. 按 README 示例运行（LLMAgent + `mcpServers` 配置）。\n\n## 简介\n\nMS-Agent 是轻量级 Agent 框架，可通过 MCP 配置接入工具；用 pip 安装后，按示例把 Streamable HTTP 的 MCP endpoint 写入配置即可跑通 LLMAgent。\n\n- **适合谁：** 想直接消费 MCP 配置、又不想自建一套 agent harness 的团队\n- **可搭配：** Python；ModelScope API（见 README）；Streamable HTTP 的 MCP endpoint\n- **准备时间：** 15–35 分钟\n\n## 实战建议\n\n- 量化信息：README 给出 `pip install ms-agent`，并提供可选的 `[research]` extra。\n- 量化信息：先用一个 MCP endpoint 调通一次工具，并记录往返延迟，作为后续回归基线。\n\n## 把它当 harness 用，而不是一次性 demo\n\nREADME 示例能跑通，但要上生产还需要护栏：\n\n1. **按环境固化 MCP 配置**（dev\u002Fstaging\u002Fprod）并纳入版本管理。\n2. **凭证与配置分离**：只用环境变量注入，不把 secret 写进仓库。\n3. **可复现**：维护一小组“金标准任务”，升级后重复跑，快速发现回归。\n\n如果目标是评测\u002F安全，建议额外加：\n\n- 输入输出日志，\n- tool-call allow-list，\n- 每任务的成本\u002F延迟预算。\n\n### FAQ\n\n**它支持 MCP 吗？**\n答：支持。README 给出了 `mcpServers` 配置并接入 LLMAgent 的示例。\n\n**如何保证可复现？**\n答：配置固定 + 金标准任务 + 记录 tool calls 与输出。\n\n**第一步做什么？**\n答：先接一个简单 MCP endpoint 跑通单任务闭环，再逐步加工具面。\n\n## 来源与感谢\n\n> Source: https:\u002F\u002Fgithub.com\u002Fmodelscope\u002Fms-agent\n> License: Apache-2.0\n> GitHub stars: 4,244 · forks: 496\n","0",[24],{"id":25,"name":26,"slug":27,"icon":28},11,"Scripts","script","📜",false,"e7d3e671d6e0960d089b13cf0c6c50b8b00981b0e86e4e6f2e46c97c76a51acc",[32,33,34],"claude_code","codex","gemini_cli","single","ms-agent",{"executes_code":29,"modifies_global_config":29,"requires_secrets":38,"uses_absolute_paths":29,"network_access":29},null,{"npm":40,"pip":41,"brew":43,"system":44},[],[36,42],"ms-agent[research]",[],[],{"commands":46,"expected_files":47},[],[20],{"asset_kind":27,"target_tools":49,"install_mode":35,"entrypoint":36,"risk_profile":50,"dependencies":51,"content_hash":30,"verification":56},[32,33,34],{"executes_code":29,"modifies_global_config":29,"requires_secrets":38,"uses_absolute_paths":29,"network_access":29},{"npm":52,"pip":53,"brew":54,"system":55},[],[36,42],[],[],{"commands":57,"expected_files":58},[],[20],{"target":33,"score":60,"status":61,"policy":61,"why":62,"asset_kind":27,"install_mode":35},29,"stage_only",[63,64,65,66,67,68,69],"target_tools includes codex","asset_kind script","install_mode single","markdown-only","policy stage_only","asset_kind script is not activated directly for Codex","trust established",{"author_trust_level":71,"verified_publisher":29,"asset_signed_hash":30,"signature_status":72,"install_count":12,"report_count":12,"dangerous_capability_badges":73,"review_status":74,"signals":75},"established","hash_only",[27],"unreviewed",[76,77],"author has published assets","content hash available",{"owner_uuid":9,"owner_name":10,"source_url":79,"content_hash":30,"visibility":19,"created_at":80,"updated_at":81},"https:\u002F\u002Ftokrepo.com\u002Fen\u002Fworkflows\u002Fms-agent-lightweight-agent-framework-with-mcp","2026-05-12 23:31:49","2026-05-14 08:16:17",[83,133,176,244],{"id":84,"uuid":85,"slug":86,"title":87,"description":88,"author_id":9,"author_name":10,"author_avatar":11,"token_estimate":12,"time_saved":12,"model_used":13,"fork_count":12,"vote_count":12,"view_count":89,"parent_id":12,"parent_uuid":13,"lang_type":15,"steps":90,"tags":91,"has_voted":29,"visibility":19,"share_token":13,"is_featured":12,"content_hash":93,"asset_kind":27,"target_tools":94,"install_mode":35,"entrypoint":95,"risk_profile":96,"dependencies":97,"verification":102,"agent_metadata":105,"agent_fit":116,"trust":118,"provenance":121,"created_at":123,"updated_at":124,"__relatedScore":125,"__relatedReasons":126,"__sharedTags":131},3092,"716fc1ad-c3a2-49ab-a810-3a139949e377","camel-multi-agent-and-tooling-library","CAMEL — Multi-Agent and Tooling Library","CAMEL is an open-source library for multi-agent systems and tools. Install `camel-ai`, then compose agents, tools, and optional web helpers in Python.",18,[],[92],{"id":25,"name":26,"slug":27,"icon":28},"e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",[32,33,34],"README.md",{"executes_code":29,"modifies_global_config":29,"requires_secrets":38,"uses_absolute_paths":29,"network_access":29},{"npm":98,"pip":99,"brew":100,"system":101},[],[],[],[],{"commands":103,"expected_files":104},[],[],{"asset_kind":27,"target_tools":106,"install_mode":35,"entrypoint":95,"risk_profile":107,"dependencies":108,"content_hash":93,"verification":113},[32,33,34],{"executes_code":29,"modifies_global_config":29,"requires_secrets":38,"uses_absolute_paths":29,"network_access":29},{"npm":109,"pip":110,"brew":111,"system":112},[],[],[],[],{"commands":114,"expected_files":115},[],[],{"target":33,"score":60,"status":61,"policy":61,"why":117,"asset_kind":27,"install_mode":35},[63,64,65,66,67,68,69],{"author_trust_level":71,"verified_publisher":29,"asset_signed_hash":93,"signature_status":72,"install_count":12,"report_count":12,"dangerous_capability_badges":119,"review_status":74,"signals":120},[27],[76,77],{"owner_uuid":9,"owner_name":10,"source_url":122,"content_hash":93,"visibility":19,"created_at":123,"updated_at":124},"https:\u002F\u002Ftokrepo.com\u002Fen\u002Fworkflows\u002Fcamel-multi-agent-and-tooling-library","2026-05-12 00:58:31","2026-05-14 08:17:43",84.91813040142924,[127,128,129,130],"topic-match","same-kind","same-target","same-author",[27,132],"scripts",{"id":134,"uuid":135,"slug":136,"title":137,"description":138,"author_id":9,"author_name":10,"author_avatar":11,"token_estimate":12,"time_saved":12,"model_used":13,"fork_count":12,"vote_count":12,"view_count":139,"parent_id":12,"parent_uuid":13,"lang_type":15,"steps":140,"tags":141,"has_voted":29,"visibility":19,"share_token":13,"is_featured":12,"content_hash":93,"asset_kind":27,"target_tools":143,"install_mode":35,"entrypoint":95,"risk_profile":144,"dependencies":145,"verification":150,"agent_metadata":153,"agent_fit":164,"trust":166,"provenance":169,"created_at":171,"updated_at":172,"__relatedScore":173,"__relatedReasons":174,"__sharedTags":175},3093,"d814e244-7701-4e98-ac46-cd4f55dbbc02","langchain-mcp-adapters-mcp-tools-in-langgraph","langchain-mcp-adapters — MCP Tools in LangGraph","Use MCP servers inside LangChain\u002FLangGraph by loading MCP tools into agents via MultiServerMCPClient, without writing one-off wrappers per server.",13,[],[142],{"id":25,"name":26,"slug":27,"icon":28},[32,33,34],{"executes_code":29,"modifies_global_config":29,"requires_secrets":38,"uses_absolute_paths":29,"network_access":29},{"npm":146,"pip":147,"brew":148,"system":149},[],[],[],[],{"commands":151,"expected_files":152},[],[],{"asset_kind":27,"target_tools":154,"install_mode":35,"entrypoint":95,"risk_profile":155,"dependencies":156,"content_hash":93,"verification":161},[32,33,34],{"executes_code":29,"modifies_global_config":29,"requires_secrets":38,"uses_absolute_paths":29,"network_access":29},{"npm":157,"pip":158,"brew":159,"system":160},[],[],[],[],{"commands":162,"expected_files":163},[],[],{"target":33,"score":60,"status":61,"policy":61,"why":165,"asset_kind":27,"install_mode":35},[63,64,65,66,67,68,69],{"author_trust_level":71,"verified_publisher":29,"asset_signed_hash":93,"signature_status":72,"install_count":12,"report_count":12,"dangerous_capability_badges":167,"review_status":74,"signals":168},[27],[76,77],{"owner_uuid":9,"owner_name":10,"source_url":170,"content_hash":93,"visibility":19,"created_at":171,"updated_at":172},"https:\u002F\u002Ftokrepo.com\u002Fen\u002Fworkflows\u002Flangchain-mcp-adapters-mcp-tools-in-langgraph","2026-05-12 01:55:09","2026-05-14 00:42:35",84.71919205351736,[127,128,129,130],[27,132],{"id":177,"uuid":178,"slug":179,"title":180,"description":181,"author_id":182,"author_name":183,"author_avatar":184,"token_estimate":185,"time_saved":12,"model_used":13,"fork_count":12,"vote_count":19,"view_count":186,"parent_id":12,"parent_uuid":13,"lang_type":15,"steps":187,"tags":188,"has_voted":29,"visibility":19,"share_token":13,"is_featured":12,"content_hash":195,"asset_kind":196,"target_tools":197,"install_mode":35,"entrypoint":198,"risk_profile":199,"dependencies":201,"verification":206,"agent_metadata":209,"agent_fit":222,"trust":229,"provenance":237,"created_at":239,"updated_at":240,"__relatedScore":241,"__relatedReasons":242,"__sharedTags":243},498,"29944683-2033-4e30-9449-6d18dffa60f5","chrome-mcp-background-proxy-fix-popups-focus-stealing-multi-29944683","Chrome MCP Background Proxy — Fix Popups, Focus Stealing & Multi-Agent Conflicts","Persistent CDP proxy + entry script that lets chrome-devtools-mcp run against your real, logged-in Chrome without the Chrome 146+ consent popup spamming on every connection, without focus stealing (Target.activateTarget \u002F Page.bringToFront are intercepted, createTarget is forced to background), and without request-ID \u002F event collisions when multiple Claude Code windows or sub-agents share one Chrome. Includes the proxy core (cdp-proxy.mjs v3), entry script, safe cleanup, pre-flight healthcheck, and a launchd-style self-healing watchdog with Feishu alerts.","4ddba6f3-0476-4a31-8a32-a355f91e76a9","henuwangkai","https:\u002F\u002Flh3.googleusercontent.com\u002Fa\u002FACg8ocL-rjptPgUuKg643YHMlh3982PAag-kpDTQgoOC8Yb7fmJAEEM=s96-c",5358,387,[],[189,194],{"id":190,"name":191,"slug":192,"icon":193},10,"MCP Configs","mcp","🔌",{"id":25,"name":26,"slug":27,"icon":28},"0ac0428df43197633f1196e8e77402f9a323092aa251a64b3a96f9250539e93c","mcp_config",[32,33,34],"SKILL.md",{"executes_code":29,"modifies_global_config":29,"requires_secrets":200,"uses_absolute_paths":29,"network_access":29},[],{"npm":202,"pip":203,"brew":204,"system":205},[],[],[],[],{"commands":207,"expected_files":208},[],[],{"asset_kind":196,"target_tools":210,"install_mode":35,"entrypoint":198,"risk_profile":211,"dependencies":213,"content_hash":195,"verification":218,"inferred":221},[32,33,34],{"executes_code":29,"modifies_global_config":29,"requires_secrets":212,"uses_absolute_paths":29,"network_access":29},[],{"npm":214,"pip":215,"brew":216,"system":217},[],[],[],[],{"commands":219,"expected_files":220},[],[],true,{"target":33,"score":223,"status":61,"policy":61,"why":224,"asset_kind":196,"install_mode":35},31,[63,225,65,66,67,226,227,228],"asset_kind mcp_config","asset_kind mcp_config is not activated directly for Codex","trust verified","verified publisher",{"author_trust_level":230,"verified_publisher":221,"asset_signed_hash":195,"signature_status":72,"last_reviewed_at":231,"install_count":12,"report_count":12,"dangerous_capability_badges":232,"review_status":233,"signals":234},"verified","2026-05-14T05:22:20+08:00",[196],"reviewed",[235,236,76,77,228],"asset has stars","asset has usage views",{"owner_uuid":182,"owner_name":183,"source_url":238,"content_hash":195,"visibility":19,"created_at":239,"updated_at":240},"https:\u002F\u002Ftokrepo.com\u002Fen\u002Fworkflows\u002Fchrome-mcp-background-proxy-fix-popups-focus-stealing-multi-29944683","2026-04-06 11:56:43","2026-05-14 05:22:20",79.08736757104724,[127,129],[27,132],{"id":245,"uuid":246,"slug":247,"title":248,"description":249,"author_id":9,"author_name":10,"author_avatar":11,"token_estimate":12,"time_saved":12,"model_used":13,"fork_count":12,"vote_count":12,"view_count":139,"parent_id":12,"parent_uuid":13,"lang_type":15,"steps":250,"tags":251,"has_voted":29,"visibility":19,"share_token":13,"is_featured":12,"content_hash":93,"asset_kind":27,"target_tools":253,"install_mode":35,"entrypoint":95,"risk_profile":254,"dependencies":255,"verification":260,"agent_metadata":263,"agent_fit":274,"trust":276,"provenance":279,"created_at":281,"updated_at":282,"__relatedScore":283,"__relatedReasons":284,"__sharedTags":285},3083,"e4780821-245b-49db-9315-ba6260689aa5","instructor-typed-structured-outputs-for-llms","Instructor — Typed Structured Outputs for LLMs","Instructor turns LLM replies into validated Pydantic models with retries. `pip install instructor`, then extract typed objects across major providers.",[],[252],{"id":25,"name":26,"slug":27,"icon":28},[32,33,34],{"executes_code":29,"modifies_global_config":29,"requires_secrets":38,"uses_absolute_paths":29,"network_access":29},{"npm":256,"pip":257,"brew":258,"system":259},[],[],[],[],{"commands":261,"expected_files":262},[],[],{"asset_kind":27,"target_tools":264,"install_mode":35,"entrypoint":95,"risk_profile":265,"dependencies":266,"content_hash":93,"verification":271},[32,33,34],{"executes_code":29,"modifies_global_config":29,"requires_secrets":38,"uses_absolute_paths":29,"network_access":29},{"npm":267,"pip":268,"brew":269,"system":270},[],[],[],[],{"commands":272,"expected_files":273},[],[],{"target":33,"score":60,"status":61,"policy":61,"why":275,"asset_kind":27,"install_mode":35},[63,64,65,66,67,68,69],{"author_trust_level":71,"verified_publisher":29,"asset_signed_hash":93,"signature_status":72,"install_count":12,"report_count":12,"dangerous_capability_badges":277,"review_status":74,"signals":278},[27],[76,77],{"owner_uuid":9,"owner_name":10,"source_url":280,"content_hash":93,"visibility":19,"created_at":281,"updated_at":282},"https:\u002F\u002Ftokrepo.com\u002Fen\u002Fworkflows\u002Finstructor-typed-structured-outputs-for-llms","2026-05-12 00:58:29","2026-05-14 00:41:39",67.71919205351736,[128,129,130],[27,132]]