[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7AHT9MHM8CV11M48XD39T5W":3,"skill-same:01M7AHT9MHM8CV11M48XD39T5W":48},{"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":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M7AHT9MHM8CV11M48XD39T5W","mantle-new-project",null,"Scaffold a complete, working Rasa Mantle project from scratch. Use when the user wants a new Mantle\u002FRasa agent, a new bot project, or asks to \"set up Rasa\". Produces every file with the catalog-pinned configuration shapes so the first validate passes.","RasaHQ\u002Frasa-community-resources","RasaHQ","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F21214473?v=4","01M7ABXR532XTM0MWW7ZK48J12","Rasa","01M7AGG0E6HH8BMYVXAEY9SHWJ",44,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources\u002Fblob\u002Fmain\u002Fstarter-pack\u002F.claude\u002Fskills\u002Fmantle-new-project\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: mantle-new-project\ndescription: >\n  Scaffold a complete, working Rasa Mantle project from scratch. Use when the\n  user wants a new Mantle\u002FRasa agent, a new bot project, or asks to \"set up\n  Rasa\". Produces every file with the catalog-pinned configuration shapes so the first\n  validate passes.\n---\n\n# Scaffold a new Rasa Mantle project\n\nAsk for (or infer): the agent's **domain**, a short **persona name**, and which\n**channels** it needs (text-only vs voice). Then create every file below —\ndo not skip any; each absence is a lint finding or a runtime surprise.\n\n## 1. pyproject.toml — the version facts are non-negotiable\n\n```toml\n[project]\nname = \"\u003Cproject-slug>\"\nversion = \"0.1.0\"\ndescription = \"\u003Cone line>\"\nreadme = \"README.md\"\nrequires-python = \">=3.11,\u003C3.13\"\ndependencies = [\n    \"rasa-pro==3.21.0.dev5\",\n    \"python-dotenv>=1.0.0\",\n]\n\n[tool.uv]\nprerelease = \"allow\"\n\n[build-system]\nrequires = [\"hatchling\"]\nbuild-backend = \"hatchling.build\"\n\n[tool.hatch.build.targets.wheel]\npackages = [\"lib\"]\n```\n\nWHY, so you can defend it: the Mantle engine ships **only on the\n`3.20.0.dev` pre-release line** — the newest *stable* rasa-pro contains no\nengine package at all, so \"latest stable\" produces a project whose imports\nfail. `prerelease = \"allow\"` is what lets uv resolve a dev pin. The\n`>=3.11` floor is required by 3.20; a lower floor fails `uv lock` with an\nerror about \"versions not supported by your dependencies\" that never\nmentions Python.\n\nIf the project needs credentials beyond `RASA_LICENSE` and `OPENAI_API_KEY`\n(e.g. `GEMINI_API_KEY`), also declare them:\n\n```toml\n[tool.rasa-catalog]\nrequired-secrets = [\"GEMINI_API_KEY\"]\n```\n\n## 2. lib\u002Fengine.py — the only file that imports the engine directly\n\n```python\n\"\"\"Engine imports, resolved in one place.\n\nRasa renamed the engine package to rasa.mantle in 3.20.0.dev1, with no alias\nfor the old name. Every tool imports from here, never from rasa.* directly,\nso the next rename is a one-file change.\n\"\"\"\nfrom __future__ import annotations\n\nfrom rasa.mantle.tools.decorator import ToolContext, tool\nfrom rasa.mantle.tools.result import ToolResult\n\nENGINE_PACKAGE = \"rasa.mantle\"\n\n__all__ = [\"ENGINE_PACKAGE\", \"ToolContext\", \"ToolResult\", \"tool\"]\n```\n\nAlso create empty `lib\u002F__init__.py` and `tools\u002F__init__.py`.\n\n## 3. agent.yml — prompt-tuning keys are TOP-LEVEL\n\n```yaml\nagent:\n  id: \u003Cproject-slug>\n  language: en\n  persona: |\n    You are \u003CName>, a concise assistant for \u003Cdomain>.\n    Ask one question at a time and keep answers short.\n    Never invent facts — always use tools for data.\n\n# TOP-LEVEL keys, siblings of `agent:`. Nested inside they parse without\n# error and are silently discarded — the engine reads them only from the top\n# level of this file.\nname: \u003CName>\ndescription: \u003Cone line, shown to the model>\n\nrules:\n  - \"Be polite, clear, and brief.\"\n  - \"Never state data that did not come from a tool.\"\n  - \"Activate a skill only when the request clearly matches it.\"\n```\n\nFor a **voice** agent, set the engines on the channel in `integrations.yml`\n(section 4), not in agent.yml. A nested `agent.voice` block is ignored by Rasa\nPro 3.21 and rejected by 3.22.\n\n## 4. integrations.yml — llm is a model-group reference\n\n```yaml\n# The orchestrator LLM is a model-group reference. Provider, model and\n# credentials live on the named group, never inline under `llm:` — the\n# inline form was removed in 3.20.0.dev6.\nllm:\n  model_group: orchestrator\n\nmodel_groups:\n  - id: orchestrator\n    models:\n      - provider: openai\n        model: gpt-4.1-mini\n        api_key: ${OPENAI_API_KEY}\n        temperature: 0.0\n\nchannels:\n  rest:\n    enabled: true\n  inspector:\n    enabled: true\n    # Voice agents only: the engines for this channel.\n    asr:\n      name: deepgram       # or a dotted path to a custom engine class\n    tts:\n      name: deepgram\n```\n\n## 5. memory.yml — project memory is tool-written\n\n```yaml\n# Project memory — facts that outlive a single skill. Readable and writable\n# by EVERY skill. Keep it to facts about the end user and the session.\n# NEVER put `llm_settable: true` here — the engine rejects it at the project\n# level. LLM-settable fields belong in skills\u002F\u003Cid>\u002Fmemory.yml.\nauthenticated:\n  type: bool\n  # Whether the caller has proven their identity this session.\n  initial_value: false\n```\n\nDocument a project field with a YAML comment, not `description:`. The model\ncannot set project fields, so it never sees their description, and Rasa Pro\n3.22 rejects one (`mantle.validation.memory.description_not_exposed`).\nSkill-scoped `llm_settable` fields keep `description:`, because the model\nreads it.\n\n## 6. Skills — at least one real skill plus a greeting\n\nCreate `skills\u002F\u003Cskill_id>\u002Fskill.md` using the **mantle-skill-authoring**\nskill's grammar (frontmatter → prose → top-level `if:` branches). Put tools\nonly that skill uses in `skills\u002F\u003Cskill_id>\u002Ftools.py`; shared tools in\n`tools\u002F` and reference them with `import_tools:` in the frontmatter.\n\n## 7. Credentials and hygiene\n\n- `.env.example` (committed): every key the project reads, values empty, one\n  comment per key saying where to get it. Start with:\n\n  ```bash\n  # Rasa Pro licence (free Developer Edition, a long JWT, one line)\n  RASA_LICENSE=\n  # LLM for routing and conversation\n  OPENAI_API_KEY=\n  ```\n\n- `.gitignore` must contain: `.env`, `.venv\u002F`, `models\u002F`, `__pycache__\u002F`,\n  `.rasa\u002F`.\n\n## 8. Makefile — the beginner's interface\n\n```makefile\n.DEFAULT_GOAL := help\n.PHONY: help env install validate train chat lint clean\n\nhelp: ## Show targets\n\t@grep -E '^[a-z-]+:.*##' $(MAKEFILE_LIST) | awk -F':.*## ' '{printf \"  make %-10s %s\\n\", $$1, $$2}'\nenv: ## Create .env from .env.example\n\t@test -f .env || cp .env.example .env\ninstall: ## Install dependencies\n\tuv sync --prerelease=allow\nlint: ## Offline consistency checks\n\tpython3 scripts\u002Flint_mantle.py\nvalidate: lint ## Validate the project without training\n\tuv run rasa validate\ntrain: ## Validate and train\n\tuv run rasa train\nchat: ## Talk to the agent in the Inspector\n\tuv run rasa inspect\nclean: ## Remove models and caches\n\trm -rf models .rasa\n```\n\n## 9. Finish\n\n1. Copy `scripts\u002Flint_mantle.py` and `hooks\u002F` from the starter pack if not\n   present; run `.\u002Fhooks\u002Finstall-hooks.sh`.\n2. Run `python3 scripts\u002Flint_mantle.py` — must be clean before the first commit.\n3. Tell the user exactly which of the two gates you ran: the offline lint\n   (consistency) vs `make validate`\u002F`train` (actually exercises the engine —\n   needs `uv sync` and a `RASA_LICENSE`). A green lint alone is not \"it works\".\n","# Scaffold a new Rasa Mantle project\n\nAsk for (or infer): the agent's **domain**, a short **persona name**, and which\n**channels** it needs (text-only vs voice). Then create every file below —\ndo not skip any; each absence is a lint finding or a runtime surprise.\n\n## 1. pyproject.toml — the version facts are non-negotiable\n\n```toml\n[project]\nname = \"\u003Cproject-slug>\"\nversion = \"0.1.0\"\ndescription = \"\u003Cone line>\"\nreadme = \"README.md\"\nrequires-python = \">=3.11,\u003C3.13\"\ndependencies = [\n    \"rasa-pro==3.21.0.dev5\",\n    \"python-dotenv>=1.0.0\",\n]\n\n[tool.uv]\nprerelease = \"allow\"\n\n[build-system]\nrequires = [\"hatchling\"]\nbuild-backend = \"hatchling.build\"\n\n[tool.hatch.build.targets.wheel]\npackages = [\"lib\"]\n```\n\nWHY, so you can defend it: the Mantle engine ships **only on the\n`3.20.0.dev` pre-release line** — the newest *stable* rasa-pro contains no\nengine package at all, so \"latest stable\" produces a project whose imports\nfail. `prerelease = \"allow\"` is what lets uv resolve a dev pin. The\n`>=3.11` floor is required by 3.20; a lower floor fails `uv lock` with an\nerror about \"versions not supported by your dependencies\" that never\nmentions Python.\n\nIf the project needs credentials beyond `RASA_LICENSE` and `OPENAI_API_KEY`\n(e.g. `GEMINI_API_KEY`), also declare them:\n\n```toml\n[tool.rasa-catalog]\nrequired-secrets = [\"GEMINI_API_KEY\"]\n```\n\n## 2. lib\u002Fengine.py — the only file that imports the engine directly\n\n```python\n\"\"\"Engine imports, resolved in one place.\n\nRasa renamed the engine package to rasa.mantle in 3.20.0.dev1, with no alias\nfor the old name. Every tool imports from here, never from rasa.* directly,\nso the next rename is a one-file change.\n\"\"\"\nfrom __future__ import annotations\n\nfrom rasa.mantle.tools.decorator import ToolContext, tool\nfrom rasa.mantle.tools.result import ToolResult\n\nENGINE_PACKAGE = \"rasa.mantle\"\n\n__all__ = [\"ENGINE_PACKAGE\", \"ToolContext\", \"ToolResult\", \"tool\"]\n```\n\nAlso create empty `lib\u002F__init__.py` and `tools\u002F__init__.py`.\n\n## 3. agent.yml — prompt-tuning keys are TOP-LEVEL\n\n```yaml\nagent:\n  id: \u003Cproject-slug>\n  language: en\n  persona: |\n    You are \u003CName>, a concise assistant for \u003Cdomain>.\n    Ask one question at a time and keep answers short.\n    Never invent facts — always use tools for data.\n\n# TOP-LEVEL keys, siblings of `agent:`. Nested inside they parse without\n# error and are silently discarded — the engine reads them only from the top\n# level of this file.\nname: \u003CName>\ndescription: \u003Cone line, shown to the model>\n\nrules:\n  - \"Be polite, clear, and brief.\"\n  - \"Never state data that did not come from a tool.\"\n  - \"Activate a skill only when the request clearly matches it.\"\n```\n\nFor a **voice** agent, set the engines on the channel in `integrations.yml`\n(section 4), not in agent.yml. A nested `agent.voice` block is ignored by Rasa\nPro 3.21 and rejected by 3.22.\n\n## 4. integrations.yml — llm is a model-group reference\n\n```yaml\n# The orchestrator LLM is a model-group reference. Provider, model and\n# credentials live on the named group, never inline under `llm:` — the\n# inline form was removed in 3.20.0.dev6.\nllm:\n  model_group: orchestrator\n\nmodel_groups:\n  - id: orchestrator\n    models:\n      - provider: openai\n        model: gpt-4.1-mini\n        api_key: ${OPENAI_API_KEY}\n        temperature: 0.0\n\nchannels:\n  rest:\n    enabled: true\n  inspector:\n    enabled: true\n    # Voice agents only: the engines for this channel.\n    asr:\n      name: deepgram       # or a dotted path to a custom engine class\n    tts:\n      name: deepgram\n```\n\n## 5. memory.yml — project memory is tool-written\n\n```yaml\n# Project memory — facts that outlive a single skill. Readable and writable\n# by EVERY skill. Keep it to facts about the end user and the session.\n# NEVER put `llm_settable: true` here — the engine rejects it at the project\n# level. LLM-settable fields belong in skills\u002F\u003Cid>\u002Fmemory.yml.\nauthenticated:\n  type: bool\n  # Whether the caller has proven their identity this session.\n  initial_value: false\n```\n\nDocument a project field with a YAML comment, not `description:`. The model\ncannot set project fields, so it never sees their description, and Rasa Pro\n3.22 rejects one (`mantle.validation.memory.description_not_exposed`).\nSkill-scoped `llm_settable` fields keep `description:`, because the model\nreads it.\n\n## 6. Skills — at least one real skill plus a greeting\n\nCreate `skills\u002F\u003Cskill_id>\u002Fskill.md` using the **mantle-skill-authoring**\nskill's grammar (frontmatter → prose → top-level `if:` branches). Put tools\nonly that skill uses in `skills\u002F\u003Cskill_id>\u002Ftools.py`; shared tools in\n`tools\u002F` and reference them with `import_tools:` in the frontmatter.\n\n## 7. Credentials and hygiene\n\n- `.env.example` (committed): every key the project reads, values empty, one\n  comment per key saying where to get it. Start with:\n\n  ```bash\n  # Rasa Pro licence (free Developer Edition, a long JWT, one line)\n  RASA_LICENSE=\n  # LLM for routing and conversation\n  OPENAI_API_KEY=\n  ```\n\n- `.gitignore` must contain: `.env`, `.venv\u002F`, `models\u002F`, `__pycache__\u002F`,\n  `.rasa\u002F`.\n\n## 8. Makefile — the beginner's interface\n\n```makefile\n.DEFAULT_GOAL := help\n.PHONY: help env install validate train chat lint clean\n\nhelp: ## Show targets\n\t@grep -E '^[a-z-]+:.*##' $(MAKEFILE_LIST) | awk -F':.*## ' '{printf \"  make %-10s %s\\n\", $$1, $$2}'\nenv: ## Create .env from .env.example\n\t@test -f .env || cp .env.example .env\ninstall: ## Install dependencies\n\tuv sync --prerelease=allow\nlint: ## Offline consistency checks\n\tpython3 scripts\u002Flint_mantle.py\nvalidate: lint ## Validate the project without training\n\tuv run rasa validate\ntrain: ## Validate and train\n\tuv run rasa train\nchat: ## Talk to the agent in the Inspector\n\tuv run rasa inspect\nclean: ## Remove models and caches\n\trm -rf models .rasa\n```\n\n## 9. Finish\n\n1. Copy `scripts\u002Flint_mantle.py` and `hooks\u002F` from the starter pack if not\n   present; run `.\u002Fhooks\u002Finstall-hooks.sh`.\n2. Run `python3 scripts\u002Flint_mantle.py` — must be clean before the first commit.\n3. Tell the user exactly which of the two gates you ran: the offline lint\n   (consistency) vs `make validate`\u002F`train` (actually exercises the engine —\n   needs `uv sync` and a `RASA_LICENSE`). A green lint alone is not \"it works\".\n",{"name":5,"description":7},"main","47c2269dde807f58a4e038be76e92da85c71060b","starter-pack\u002F.claude\u002Fskills\u002Fmantle-new-project",19,"https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources","https:\u002F\u002Frasa.community\u002F","https:\u002F\u002Fgithub.com\u002FRasaHQ","https:\u002F\u002Frasa.com","Rasa is a conversational AI framework for building reliable, scalable AI agents.","2026-11-14 12:02:45",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add RasaHQ\u002Frasa-community-resources --skill mantle-new-project",{"tool":40,"command":41},"bunx","bunx skills add RasaHQ\u002Frasa-community-resources --skill mantle-new-project",{"tool":43,"command":44},"pnpm","pnpm dlx skills add RasaHQ\u002Frasa-community-resources --skill mantle-new-project","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources\u002Fblob\u002Fmain\u002Fstarter-pack\u002F.claude\u002Fskills\u002Fmantle-new-project\u002FSKILL.md",true,[],[49,56,63,70,77,82],{"skillNo":50,"name":51,"nameZh":6,"description":52,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":53,"categories":54,"keywords":55,"publishedAt":19},"01M7AJ24E15F9CMBGM3QZSKV6S","Build a travel desk","Handle developer requests to build or extend a Mantle travel agent using four approved travel tools.","https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources\u002Fblob\u002Fmain\u002Fworkshops\u002Fsoftware-factories-masterclass\u002Fstudio\u002Fskills\u002Fbuild_agent\u002Fskill.md",[],[],{"skillNo":57,"name":58,"nameZh":6,"description":59,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7AJ20RB027VHFWF2MAM79WC","Travel briefing","Answer questions about flight status, tonight’s departures, airport weather and currency conversion.","https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources\u002Fblob\u002Fmain\u002Fworkshops\u002Fsoftware-factories-masterclass\u002Freference\u002Ffinished-agent\u002Fskills\u002Ftravel_briefing\u002Fskill.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7AJ1XQ9HR7H1ENZRX9YF5EJ","Review Dispute","Ask the outside reviewer about a card dispute, then block the card if the caller confirms. Activate when the caller reports a card transaction they did not make.","https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources\u002Fblob\u002Fmain\u002Fworkshops\u002Fsoftware-factories-masterclass\u002Fopenai-mcp\u002Fproject\u002Fskills\u002Freview_dispute\u002Fskill.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7AJ1W0KTP19WYFACXY9WTRD","Block Card","Block a customer's lost or stolen card. Activate when the caller wants a card blocked.","https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources\u002Fblob\u002Fmain\u002Fworkshops\u002Fsoftware-factories-masterclass\u002Ffactory\u002Fcandidates\u002Fblock-card-tamper\u002Fproduct\u002Fskills\u002Fblock_card\u002Fskill.md",[],[],{"skillNo":78,"name":72,"nameZh":6,"description":73,"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},"01M7AJ1T5489X484B6V0KPR2AA","https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources\u002Fblob\u002Fmain\u002Fworkshops\u002Fsoftware-factories-masterclass\u002Ffactory\u002Fcandidates\u002Fblock-card-hangs\u002Fproduct\u002Fskills\u002Fblock_card\u002Fskill.md",[],[],{"skillNo":83,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":19},"01M7AJ1RBPS99DXH3YY373ERWW","https:\u002F\u002Fgithub.com\u002FRasaHQ\u002Frasa-community-resources\u002Fblob\u002Fmain\u002Fworkshops\u002Fsoftware-factories-masterclass\u002Ffactory\u002Fcandidates\u002Fblock-card-good\u002Fproduct\u002Fskills\u002Fblock_card\u002Fskill.md",[],[]]