菜单
Stand up a full-stack Claude-powered WhatsApp bot from the chatbot-toolkit reference app — FastAPI webhook, Channel/Brain/SessionStore/Guardrails modules, end to end. Use when starting a new chatbot, scaffolding a messaging bot, or wiring Claude to WhatsApp/Instagram.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | scaffold-bot |
| description | Stand up a full-stack Claude-powered WhatsApp bot from the chatbot-toolkit reference app — FastAPI webhook, Channel/Brain/SessionStore/Guardrails modules, end to end. Use when starting a new chatbot, scaffolding a messaging bot, or wiring Claude to WhatsApp/Instagram. |
Start from the bundled reference app — one runnable WhatsApp bot you copy and adapt.
cp .env.example .env # ANTHROPIC_API_KEY, META_* , WHATSAPP_*
uv sync
uv run uvicorn app.main:create_default_app --factory --port 8000
npx cloudflared tunnel --url http://localhost:8000 # public HTTPS for Meta
The reference app lives at reference-app/ in this plugin. Copy it into your repo
and adapt.
Four deep modules behind simple interfaces, wired by a thin webhook:
| Module | Interface | Swap for tests |
|---|---|---|
Channel | verify / verify_signature / parse / send | mock httpx transport |
Brain | respond(history, incoming) -> str | stub brain |
SessionStore | get_history / append | in-memory impl |
Guardrails | screen_input / screen_output | no-op impl |
Flow: parse → guard → load → brain → append → send. The FastAPI layer is thin
glue; logic lives in the modules. Build the app via create_app(channel, brain, store, guardrails) so you can inject fakes.
bot-session-statebot-safetymeta-messagingdeploy-botTeach the Claude Agent SDK (claude-agent-sdk) and how to layer it behind the reference app's Brain interface — the agent loop, in-process tools, MCP, and context management. Use when a bot needs an agent that runs tools/files/MCP autonomously, when outgrowing a hand-rolled messages.create loop, or when upgrading ClaudeBrain without changing respond().
Teach the raw Anthropic Python SDK behind the reference app's Brain — messages.create, the tool-use loop, streaming, and prompt caching. Use when implementing or extending a bot Brain, calling Claude directly, adding tools/streaming/caching to messages.create, or asking how ClaudeBrain talks to the model.
Make a public Claude bot safe with deterministic guardrails — block prompt-injection on input, redact PII (email/phone/SSN/card) on input and output, and short-circuit blocked messages with a canned reply. Use when hardening a bot for public traffic, adding input/output screening, or filling the chatbot-toolkit Guardrails seam.
Persist bot conversation history in Postgres using the chatbot-toolkit reference app's PostgresSessionStore — asyncpg, a lazy pool, and a temp-Postgres test. Use when sessions must survive restarts, moving a bot off in-memory storage, or adding a durable SessionStore.
Get the chatbot-toolkit reference app onto a public HTTPS endpoint Meta will accept — build the bundled Dockerfile and deploy to a container PaaS (Fly / Railway / Render) with env/secret config, or tunnel to your laptop for dev. Use when deploying a bot, Dockerizing the reference app, setting up a webhook URL Meta accepts, configuring deploy secrets, or exposing a local bot for webhook testing.
Connect the chatbot-toolkit reference app to Meta — create a Meta App, pass the webhook verify handshake, validate X-Hub-Signature-256, and send via the Meta Graph API for WhatsApp and Instagram. Use when setting up a Meta App, wiring a WhatsApp or Instagram webhook, debugging webhook verification or signature failures, parsing inbound payloads, or sending WhatsApp templates/media or Instagram DMs.