本文へ移動
cccskills

「dd」の検索結果

9,847 件 ・ 関連度順

概要と使いどころ

Integrate with the FastAPI backend through the typed wrapper in frontend/lib/api.ts, attach Supabase JWTs for admin calls, and handle errors via ApiError. Use when adding/modifying API calls or wiring auth.

日本語の概要は準備中です。原文の説明を表示しています。

shanraisshan/learning-x-claude22026年5月7日 更新

Build FastAPI endpoints in the layered router → service → Supabase pattern, with Pydantic v2 schemas, JWT/admin gating, and consistent HTTPException-based domain errors. Use when adding/modifying any endpoint under backend/.

日本語の概要は準備中です。原文の説明を表示しています。

shanraisshan/learning-x-claude22026年5月7日 更新

Complete security architecture overhaul for claude-flow v3. Addresses critical CVEs (CVE-1, CVE-2, CVE-3) and implements secure-by-default patterns. Use for security-first v3 implementation.

日本語の概要は準備中です。原文の説明を表示しています。

ruvnet/mcp-studio22026年10月10日 更新

Build production-grade AI agents from scratch using patterns inspired by production-grade agentic systems like Claude Code. Covers two audiences — (1) workflow builders who need planner+executor pipelines, human-in-the-loop checkpoints, and multi-agent orchestration, and (2) application embedders who need HTTP/SSE APIs, multi-user session management, and React frontend integration. Use when asked to build an AI agent, create a coding assistant, embed an agent in an app, build an API agent, clone Claude Code, build a CLI agent, design an agent architecture, scaffold an agentic system, or automate a multi-step workflow. Triggers on keywords like agent builder, agent blueprint, coding assistant, CLI agent, agentic tool, claude code clone, agent scaffold, agent architecture, workflow automation, multi-agent, planner executor, SSE streaming, agent API.

日本語の概要は準備中です。原文の説明を表示しています。

OthmanAdi/agent-blueprint22026年4月6日 更新

Convert Claude's Markdown output to each channel's native text syntax before delivery. Adds zero-dependency formatting for WhatsApp, Telegram, and Slack (marker substitution). Also ships a Signal rich-text helper (parseSignalStyles) used by the Signal skill.

日本語の概要は準備中です。原文の説明を表示しています。

nanocoai/nanoclaw-gmail22026年4月3日 更新

browsing

無料

Tiered web access router. TIER 1 insane-search: headless extraction plus WAF bypass via curl_cffi, yt-dlp, Jina Reader, public APIs, Playwright fallback for blocked/403/Cloudflare pages, articles, YouTube subtitles. TIER 1.5 agent-reach: platform-native readers for Chinese/social platforms: Xiaohongshu, Douyin, Weibo, Bilibili, V2EX, WeChat, plus Twitter/Reddit/LinkedIn/GitHub. TIER 2 aside-browser FIRST (user's real logged-in browser: aside exec agent, aside repl Playwright), then CloakBrowser + agent-browser as the stealth fallback: source-patched stealth Chromium (CloakBrowser) driven over CDP by the agent-browser CLI for clicks, forms, screenshots, video, cookie login, QA/dogfood, Electron apps. Triggers: browse, stealth browser, CloakBrowser, agent-browser, cloak launch, bypass bot detection, blocked site, scrape, fill form, screenshot, import cookies, play youtube, podcast transcript. NOT for simple searches (web-search), fetches (webfetch), or messaging/workspace platforms — Slack/Discord/Notion/Gmail have Tier-1 CLIs in external-communication; come here only as its fallback.

日本語の概要は準備中です。原文の説明を表示しています。

code-yeongyu/browsing-skill22026年7月12日 更新

Gates whether a new test should exist and forces it to be efficient, protecting CI from low-value test bloat. Use before any change to what a pytest, Jest, or Playwright test asserts or sets up, down to one fixture or one assertion added to an existing block. Front-loads the value bar (every test must catch a realistic regression no existing test already catches; extend the nearest existing test before writing a new standalone one; test behavior through the public interface, not implementation details; collapse near-duplicates into parameterized cases) and the efficiency bar (deterministic, isolated, fast; pick the cheapest test level; Django TestCase over TransactionTestCase; no sleeps, no real network; no time bombs from absolute dates left to age against the real clock; no database a test never uses). Includes a "don't write it" decision tree. For fixing an existing flaky test use `/fixing-flaky-tests`; after this gate says a Playwright test is warranted, use `/playwright-test` for mechanics.

日本語の概要は準備中です。原文の説明を表示しています。

amplifthq/opentag02026年10月14日 更新