Guides developers through hands-on TRTC integration from first setup to working code. Use when the user expresses intent to integrate, build, add a feature, or run a demo (e.g. "get started", "帮我接入", "从零开始", "I want to add X", "implement X", "try the demo"); or troubleshoot an issue ("报错", "crash", "not working", "黑屏"); or when the project has no TRTC dependencies detected; or when the user describes a real-world business scenario that implies real-time audio/video or messaging (e.g. telemedicine, online education, video interview, live commerce, customer service). Provides demo quickstart, scenario integration, single-feature integration, and error diagnosis.
Helps developers integrate and troubleshoot Tencent Real-Time Communication SDKs (Chat, Call, RTC Engine, Live, Conference) across Web, Android, iOS, Flutter, and Electron. Use when the user discusses real-time audio/video, live streaming, video conferencing, instant messaging, or voice/video calling scenarios; or mentions specific products like TRTC, Chat, Call, RTC Engine, Live, Conference, or TRTC error codes (6206, 6208, 70001); also when TRTC imports or class names appear in code without explicit mention of "TRTC". Also triggers on real-world business scenarios that inherently require real-time audio/video or messaging capabilities — e.g. telemedicine, online education, video interview, live commerce, customer service. Handles integration guidance, factual lookups, scenario walkthroughs, and error diagnosis.
INTERNAL skill — only invoked by trtc-topic during scenario execution. Do NOT trigger directly from user messages. Generates room-type audio/video product UI (conference, live room, voice chat) using templates, component library, or official RoomKit integration. Always route user requests through trtc-onboarding first.
Answers factual, conceptual, and decision-making questions about TRTC from authoritative sources. Use when the user asks about pricing, quotas, error codes, API usage, product comparisons, migration, or how something works conceptually — in any phrasing (e.g. "how much does X cost", "多少钱", "X vs Y", "对比", "error code 6206", "错误码", "does TRTC support Y", "配额", "the correct way to use X", "X是什么", "how does X work", "migrate from V3 to V4", "迁移"). Provides cited answers from knowledge-base slices and official trtc.io documentation — never training-data synthesis.
Guide users step-by-step through complete TRTC integration scenarios. Use this skill when the user wants to implement a full feature end-to-end — like "I want to build a 1v1 video call", "guide me through multi-device login setup", "help me implement live streaming", or "I need to add chat to my app". Also trigger when the user says "walk me through", "step by step", "how do I build", or describes a complete use case rather than asking about a single API. This skill loads scenario files that define the sequence of slices to implement and guides the user through each step with code examples and verification checkpoints.
Evaluates the quality of the trtc skill and its knowledge base by running AI-generated code against a curated eval set and scoring both static constraint compliance and runtime SDK behavior. Use when the user mentions 跑分, 评测, eval, benchmark, 回归, quality check, or asks to compare skill versions. Internal-only tool for TRTC knowledge maintainers.
INTERNAL quality gate for TRTC code generated by onboarding / topic skills. Not a user-facing skill. Runs the full verification pipeline on AI-generated code before it is declared "done": constraint compliance (slice ALWAYS/NEVER + MUST/MUST NOT rules) → compilation (xcodebuild / gradlew / npm build / flutter build) → integration safety (file-change scope, SDK init conflicts, regression tests). Triggered only by other skills in this repo (onboarding A2-Q3, topic step gates). Do NOT route to this skill in response to a user asking "review my code" or "check this implementation" — those requests are not part of this product's public interface.
Discovers matching slices from the TRTC knowledge base given a natural-language query. Use when onboarding or docs needs to find slice IDs for a feature description, symptom, error code, or pattern query. Accepts (product, platform, query, intent) and returns matched slices with confidence and source attribution. Not user-facing — callers compose the final answer from returned results.