@tencent-rtc/trtc-agent-skills 0.1.3 → 0.1.4

package/skills/trtc-onboarding/reference/path-b-troubleshoot.md

@@ -1,186 +0,0 @@
-# Path B — Troubleshooting
-
-> Loaded by `../../trtc-onboarding/SKILL.md` when `intent = troubleshoot` in `${CLAUDE_PROJECT_DIR}/.trtc-session.yaml`, OR when the user's message matches review / audit / cross-check / validate / 帮我看看 / 是否正确 (Hard rule #1 triggers B-Q0).
-> Before reading this file, SKILL.md must have populated `${CLAUDE_PROJECT_DIR}/.trtc-session.yaml` and passed Stage 1 calibration.
-
-**Your role: Debugger or integration auditor** — depending on B-Q0 classification. Walk the right tree; do not default to code-style review.
-
-Recap examples (translate to user's language):
-- Runtime: *Got it — something's broken in your Conference Web / RoomKit integration. Let me narrow down the symptom.*
-- Audit: *You want to check whether your meeting flow matches the official integration path — I'll walk the audit checklist.*
-
----
-
-## B-Q0 — Intent triage (BEFORE B-Q1)
-
-**Trigger**: ANY of:
-- `intent = troubleshoot` in session
-- review / audit / cross-check / validate / 审查 / 帮我看看 / 是否正确 / check my X
-- 检查遗漏 / 业务流程 / 对照官方 / FAQ / 排障
-
-**Core idea**: triage to the **underlying intent**. Do NOT refuse; do NOT produce code-style review.
-
-### Step 1 — Self-classify (6 intents)
-
-| Intent | Signal | Route |
-|--------|--------|-------|
-| **A. Runtime symptom** | 报错 / 崩溃 / 黑屏 / 无声音 / 进房失败 / doesn't work / login fails / 不工作 + concrete failure | **B-Q1** symptom tree (product-aware) |
-| **B. Error code** | Numeric error code (6206, 100211, 100006, 70001, …) | `docs` skill (`intent=slice-lookup`, search `intent=error-code`) |
-| **C. Official pattern** | 官方推荐 / 标准接入 / 应该怎么接 / correct pattern / host create-join-leave sequence | `docs` skill (`intent=slice-lookup`, search `intent=pattern`) |
-| **D. API comparison** | X vs Y / X 还是 Y / 两套检测怎么选 / 预制 UI vs State API | `docs` skill (`intent=slice-lookup`, search `intent=pattern`) |
-| **F. Integration audit** | 检查遗漏 / 业务流程是否正常 / 对照官方流程 / 有没有漏 / integration checklist / 在线课堂流程 / audit 项目 | **B-Audit** (read `conference/integration-audit` slice — no B-Q1) |
-| **E. Pure style review** | 写得怎么样 / 命名对不对 / 代码质量 — no symptom, error, pattern, or checklist ask | Decline (Step 3.E) |
-
-**F vs C disambiguation**:
-- **F** = user wants a **checklist against their project or scenario** ("我的流程完整吗").
-- **C** = user wants **how to implement** one official pattern ("标准 login/join 顺序是什么") with no project audit ask.
-
-If **F** and **A** both appear ("检查进房失败流程有没有遗漏"), prefer **A** when a concrete failure is described; prefer **F** when the ask is purely completeness.
-
-If high confidence → skip Step 2.
-
-### Step 2 — If ambiguous, ask ONE triage question
-
-> Which is closest?
->
-> 1. Something is broken right now — I want to fix a runtime issue
-> 2. I have an error code to look up
-> 3. I want the official recommended integration pattern for X
-> 4. I want to compare two APIs or approaches (X vs Y)
-> 5. I want to audit my meeting / classroom flow against the official checklist
-> 6. I just want code style feedback
-> 7. Type something
-
-(`AskUserQuestion`: max 4 explicit options — split into two questions if needed: first "runtime vs audit vs docs", then sub-choice.)
-
-### Step 3 — Route
-
-- **A** → B-Q1
-- **B, C, D** → hand off to `../../trtc-docs/SKILL.md` with appropriate `intent`
-- **F** → B-Audit below
-- **E** → decline:
-  > Code-style review isn't a standalone service here. If you have a concrete problem (options 1–5), I can help directly.
-
-### Step 4 — Answer-shape constraints
-
-See **Hard rule #1** in `SKILL.md`. Forbidden: Critical Review Checklist, ✅ vs ❌ main structure, "改进建议" lists, "Fixed version of your code" as the deliverable.
-
-**Allowed for F (audit)**: checklist with ✅/⚠️ per **integration item** (not per code line), slice citations, "可能遗漏" prioritized list.
-
-### Step 5 — Relapse guard
-
-If user says "帮我诊断 / go ahead", stay in the branch B-Q0 assigned — do not revert to review shape.
-
----
-
-## B-Audit — Integration audit (intent F)
-
-**Slice**: Read `${CLAUDE_PLUGIN_ROOT}/knowledge-base/slices/conference/web/integration-audit.md` (`conference/integration-audit`) and follow its §7 output template.
-
-**Workflow**:
-1. Read session: `product`, `platform`, `ui_mode`, `scenario`.
-2. If user project is available, scan for TRTC entry points (`conference.login`, `joinRoom`, `PreConferenceView`, `useDeviceState`, etc.) — **only to tick checklist items**, not to critique style.
-3. Present audit using slice §7 template. Cite slice IDs internally; user-facing text references capability names + official doc links.
-4. **Do NOT** run Fix delivery (no code patch unless user then picks a specific runtime symptom → return to B-Q1).
-
-**Direct triggers** (skip B-Q1 even without "review" wording):
-- 检查…遗漏 / 业务流程 / 对照官方 / 在线课堂流程 / RoomKit 接入要点 / integration checklist
-
----
-
-## B-Q1 — Symptom tree (product-aware)
-
-After intent **A**, pick the menu from session `(product, platform)`. Default to **Conference Web** when session says `(conference, web)` OR when TRTC deps in `package.json` include `@tencentcloud/roomkit-web-vue3` / `tuikit-atomicx-vue3`.
-
-### B-Q1a — Conference Web (default for v1)
-
-Question: *Which best matches what you're seeing?*
-
-| # | Option | Loaded trees | Context to request inline |
-|---|--------|--------------|---------------------------|
-| 1 | Can't enter room / room not found / dismissed / wrong password | `conference/login-auth` + `conference/room-lifecycle` | login done? roomId source? join vs createAndJoin? |
-| 2 | No video / no audio / permission denied / device busy | `conference/prejoin-check` + `conference/device-control` | HTTPS? iframe? prejoin vs in-meeting API mix? |
-| 3 | Login / UserSig / 6206 / kicked offline / session expired | `conference/login-auth` | userSig source? onLoginExpired handler? |
-| 4 | Pre-join device check conflicts with in-meeting state (dual getUserMedia / DeviceDetector) | `conference/prejoin-check` + `conference/device-control` | which APIs used for prejoin vs in-room? |
-| 5 | Screen share fails / iframe blocked / display-capture | `conference/screen-share` + `conference/device-control` | browser? iframe allow attr? |
-| 6 | Room type / capability mismatch (100211, 100006, live tools in standard room) | `conference/integration-audit` §6 + `conference/room-lifecycle`; then `docs` if code unknown | roomType? Live APIs called? |
-| 7 | Layout / remote video not updating / grid issues | `conference/video-layout` + `conference/participant-list` | official-roomkit vs custom useRoomView? |
-| 8 | Specific error code (paste it) | search `intent=error-code` | wait for code |
-| 9 | Type something | search `intent=troubleshoot` | free-text |
-
-**Conference Web quick fixes reference** (when no dedicated error-codes slice match):
-
-| Code | Verdict | Action |
-|------|---------|--------|
-| 6206 | UserSig invalid/expired | Re-issue userSig; listen `USER_SIG_EXPIRED` |
-| 100211 | take-seat not enabled | Standard room: usually **safe to ignore** if not using seat/co-guest APIs |
-| 100006 | live-only API on standard room | Remove Live/PK/co-host calls; hide Live widgets via `setWidgetVisible` |
-
-### B-Q1b — Live iOS (legacy menu)
-
-Use when `(product, platform) == (live, ios)`:
-
-| # | Option | Loaded tree |
-|---|--------|-------------|
-| 1 | Black screen / video not rendering | anchor-preview + audience-watch |
-| 2 | Crash / app freezes | anchor-lifecycle cleanup |
-| 3 | Specific error code | search `intent=error-code` |
-| 4 | Audio but no video (or vice versa) | device-control |
-| 5 | Connection fails / can't enter room | login-auth + anchor-lifecycle |
-| 6 | UI layout broken | coguest-apply layout |
-| 7 | Type something | search `intent=troubleshoot` |
-
-### B-Q1c — Other (product, platform)
-
-Question text: describe the symptom in free text → always call `../../trtc-search/SKILL.md` with `intent=troubleshoot`, `product`, `platform` from session.
-
-### Free-text / option "Type something" handling
-
-Call `../../trtc-search/SKILL.md` with `intent=troubleshoot`. Read `response` per Response Contract:
-
-- `matched` → walk `matches[].file_paths_read` troubleshooting sections
-- `status_planned` → index description + fall back to `docs`
-- `no_match` / `no_slice` → `docs`
-- `ambiguous_product` → ask user, re-call search
-
-### Code access (all B-Q1 variants)
-
-After symptom pick:
-- If `project_state` scanned → diagnose against code; no extra "how to share" question.
-- Else → inline one context request in the first diagnostic message (e.g. paste `joinRoom` call site).
-
----
-
-## B-Q2 — (merged into B-Q1)
-
-*Intentionally absent.*
-
----
-
-## Fix-write support gate
-
-Diagnosis runs for **any** (product, platform). Fix **code** only when:
-
-- `(product, platform) == (conference, web)` → full Fix delivery steps 1–4
-- Otherwise → steps 1–3 only + official docs link from llms.txt
-
-**B-Audit (intent F)** never auto-writes fix code unless user escalates to a specific runtime symptom.
-
-Source: `reference/supported-matrix.md` → Troubleshoot path.
-
----
-
-## Fix delivery
-
-When root cause identified (intent A path):
-
-1. Explain **why** (one sentence).
-2. Show **fix** (diff or corrected snippet).
-3. Cite slice ALWAYS/NEVER violated.
-4. Apply change if file access, else paste-ready block.
-
----
-
-## No verification question
-
-Do not ask "did it work?" after fix. User returns with new symptom → B-Q1 again.

package/skills/trtc-onboarding/reference/path-c-expand.md

@@ -1,43 +0,0 @@
-# Path C — Feature Expansion
-
-> Loaded by `../../trtc-onboarding/SKILL.md` when `intent = expand` in `${CLAUDE_PROJECT_DIR}/.trtc-session.yaml` (user explicitly said "my existing project already has X, now add Y / 已经接了 X，现在想加 Y").
-> Before reading this file, SKILL.md must have populated `${CLAUDE_PROJECT_DIR}/.trtc-session.yaml` and passed Stage 1 calibration.
-
-**Your role: Advisor + Implementer.**
-
-## C.1 — Auto-detect existing setup
-
-If file scanning is available, identify which TRTC features are integrated by scanning for Store class usage (`LoginStore`, `DeviceStore`, `LiveListStore`, `BarrageStore`, `GiftStore`, etc.) and populate `project_state.existing_features`.
-
-Present findings as a recap:
-
-> I can see you already have:
-> - Login (LoginStore)
-> - Live streaming (LiveCoreView + LiveListStore)
-> - Device control (DeviceStore)
->
-> Not yet integrated: Barrage, Gift, Co-guest, Beauty, Audio effects.
-
-## C-Q1 — Which feature to add
-
-Question text: "Which feature do you want to add next?"
-
-Show only the unintegrated features for the identified product, plus `Type something`. Layout and slice resolution match A2-Q1 (see `reference/path-a2-integrate.md`) but as single-select.
-
-## C.2 — Implementation
-
-After the user picks a feature, delegate to Path A2's step-by-step implementation flow (A2-Q3 per-step confirmation, including the internal apply quality gate). See `reference/path-a2-integrate.md` for the full contract.
-
-## C-Q2 — Suggest related
-
-After the feature is integrated (and has silently passed the internal compile gate), based on `cross_product_relations` and scenario co-occurrence:
-
-Question text: "Related features that often pair well with this one:"
-
-| # | Option |
-|---|--------|
-| 1-3 | Context-specific suggestions (e.g., after barrage → gift / audience-manage / co-guest) |
-| 4 | I'm good for now |
-| 5 | Type something |
-
-Picking 1-3 loops back to C.2 implementation. Picking 4 ends onboarding cleanly.

package/skills/trtc-onboarding/reference/supported-matrix.md

@@ -1,100 +0,0 @@
-# TRTC Skill — Supported Matrix (一期)
-
-> Single source of truth for which (product × platform × scenario)
-> combinations the skill currently covers, broken down by path. Update
-> this file when adding new platforms, products, or scenarios; no other
-> file should encode per-pair support state.
-
-## Integration path — write code into user project
-
-> Used by onboarding Path A2 (intent = integrate-scenario / integrate-feature)
-> and Path C (expand). Onboarding hard-blocks combinations marked ⏳; topic
-> assumes any session reaching it has already passed all three gates
-> (product, platform, scenario).
-
-### Product × Platform
-
-| Product    | Web | iOS | Android | Flutter | Electron |
-|------------|:---:|:---:|:-------:|:-------:|:--------:|
-| Conference | ✅  | ⏳  | ⏳      | ⏳      | ⏳       |
-| Live       | ⏳  | ⏳  | ⏳      | ⏳      | ⏳       |
-| Call       | ⏳  | ⏳  | ⏳      | ⏳      | ⏳       |
-| Chat       | ⏳  | ⏳  | ⏳      | ⏳      | ⏳       |
-| RTC Engine | ⏳  | ⏳  | ⏳      | ⏳      | ⏳       |
-
-✅ supported · ⏳ coming soon
-
-### Conference Web — Scenarios
-
-Within (Conference, Web), only these scenarios are integration-supported in v1:
-
-| Scenario id                        | Display name           | Support |
-|------------------------------------|------------------------|:-------:|
-| `general-conference`               | 通用会议               | ✅      |
-| `1v1-video-consultation`           | 1v1 视频问诊（医疗）   | ✅      |
-| `webinar-conference`               | 研讨会 / 宣讲会        | ⏳      |
-| `medical-multidoctor-consultation` | 医疗会诊（多医生）     | ⏳      |
-
-A2-Q0's conference scenario menu MUST list only the ✅ rows above. Other
-`status: active` scenarios in `index.yaml` are hidden from the user-facing
-menu in v1 — they remain in the index as content, but onboarding does not
-expose them as integration entry points.
-
-## Demo path — clone & run official demo
-
-All combinations supported. Demo source is the official open-source repo;
-no skill-side code generation, so platform / product / scenario matrix
-does not apply.
-
-## Docs path — fact / decision / error-code lookups
-
-All combinations supported. Backed by remote llms.txt
-(`https://trtc.io/llms/{product}/{platform}.txt` — resolve platform via
-§ "llms.txt platform identifiers" below) and slice files where available;
-falls back to official trtc.io docs when slice is missing.
-
-## Troubleshoot path — error diagnosis
-
-- **Diagnose** (B-Q0 / B-Q1, root cause analysis): all combinations.
-- **Write fix code** (Path B Step 4 — apply diff or full corrected code):
-  ✅ Conference Web only (any scenario). Other combinations get root cause +
-  slice rule citation + official docs link; no Edit/Write to user files.
-
-  Note: scenario gating does not apply here — fix-write is gated on
-  (product, platform) only. Fixes are slice/error-level edits, not
-  scenario-level scaffolding.
-
-## llms.txt platform identifiers
-
-> Maps onboarding's canonical platform name → trtc.io/llms/ URL path segment.
-> Used by Path A1 (demo) and trtc-docs when constructing llms.txt fetch URLs.
-> When a cell shows multiple values, the skill MUST ask the user to choose
-> before fetching.
-
-| Product    | `web`                        | `ios`   | `android` | `flutter` | `electron` |
-|------------|:----------------------------:|:-------:|:---------:|:---------:|:----------:|
-| Conference | `web`                        | `ios`   | `android` | `flutter` | `electron` |
-| Chat       | **`react`** ★, `vue` (国内站UI) | `ios`   | `android` | `flutter` | —          |
-| Call       | `web`                        | `ios`   | `android` | `flutter` | `electron` |
-| Live       | `web`                        | `ios`   | `android` | `flutter` | `electron` |
-| RTC Engine | `web`                        | `ios`   | `android` | `flutter` | `electron` |
-
-★ = recommended default (国际站 UI, Tencent-RTC org repo)
-
-**Rules:**
-- If the cell is a single value identical to the column header → no transform needed.
-- If the cell shows multiple values → present a framework selection question to the user before proceeding. Put the ★ option first and mark it "(Recommended)".
-- If the cell is `—` → that (product, platform) combo has no llms.txt entry; fall back to the product-level `{product}.txt`.
-
-**Chat Web note:** The React demo (`github.com/Tencent-RTC/TUIKit_React`) produces the international UI ("Tencent Cloud | Chat"). The Vue demo (`github.com/TencentCloud/chat-uikit-vue`) produces the domestic Chinese UI ("腾讯云即时通信IM"). When intent=demo, default to React unless the user explicitly asks for Vue.
-
-## When to update this file
-
-- A new platform's Conference slice content lands → flip the platform cell
-  to ✅ and remove the integration-intent narrowing in Stage 1B Q3 for that
-  platform.
-- A new conference scenario gets full coverage → add a ✅ row to the
-  Scenarios table and add it to A2-Q0's conference menu in
-  `path-a2-integrate.md`.
-- A new product launches with full slice content for some platform → flip
-  that cell.

package/skills/trtc-search/SKILL.md

@@ -1,228 +0,0 @@
----
-name: trtc-search
-description: >
-  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.
----
-
-# TRTC Knowledge Base Search
-
-You search the TRTC knowledge base to find relevant atomic capabilities (slices) and integration scenarios. You handle multi-product, cross-platform searches with a structured strategy chain and return a machine-readable response.
-
----
-
-## Response Contract (read this first)
-
-Every call returns exactly this shape. The calling skill (`docs` / `onboarding`) reads fields by name — do NOT improvise the format, do NOT collapse fields into prose.
-
-```yaml
-response:
-  status: matched | no_match | no_slice | ambiguous_product | status_planned
-  matches:                             # populated when status ∈ {matched, status_planned}; empty otherwise
-    - slice_id: <product/ability>      # e.g. live/barrage
-      confidence: high | medium | low
-      match_strategy: exact | tag | product-keyword | cross-related | fuzzy
-      content_loaded: full | summary | index-only
-      file_paths_read: [...]           # actual file paths opened this turn (relative to repo root)
-      related: [...]                   # one-hop related slice_ids NOT already in matches
-  ambiguous_candidates: [...]          # populated ONLY when status = ambiguous_product, e.g. [live, chat]
-  reason: "..."                        # populated ONLY when status ≠ matched; one-sentence cause
-```
-
-### status semantics
-
-| status | Meaning | matches | ambiguous_candidates | reason |
-|--------|---------|---------|----------------------|--------|
-| `matched` | ≥ 1 slice found with usable content | non-empty | — | — |
-| `status_planned` | slice exists in index but content not yet written | non-empty, `content_loaded: index-only` | — | "slice `<id>` is status: planned" |
-| `no_slice` | product exists but has zero slices in KB (e.g., Call, Conference at early stage) | empty | — | "product `<id>` has no slices in KB" |
-| `no_match` | search ran to exhaustion without finding relevant slices | empty | — | one-line cause (e.g., "error code 99999 not in any slice.error_codes or body text") |
-| `ambiguous_product` | product was null AND fuzzy-only hits span multiple products | empty | non-empty (e.g., `[live, chat]`) | "query plausibly matches multiple products; ask user" |
-
-### confidence → match_strategy mapping
-
-Confidence is derived from match_strategy, not judged separately:
-
-- `high` ← `exact` or `tag`
-- `medium` ← `product-keyword` or `cross-related`
-- `low` ← `fuzzy`
-
-Callers should use `confidence` for quick guards (e.g., `if response.matches[0].confidence == 'high': answer from slice, skip llms.txt`) and `match_strategy` for traceability when debugging.
-
-### What callers do with each status
-
-- `matched` → read `matches[].file_paths_read`, compose answer from slice contents.
-- `status_planned` → show the slice's index-level description; fall through to caller's fallback (docs → llms.txt).
-- `no_slice` / `no_match` → caller falls back to its own escape route (docs: llms.txt lookup; onboarding: ask user / degrade).
-- `ambiguous_product` → caller asks the user which of `ambiguous_candidates` they mean, then re-calls search with the confirmed product.
-
----
-
-## Inputs (from calling skill: `docs` or `onboarding`)
-
-- **product**: Identified product (chat/call/rtc-engine/live/conference), or `null` if ambiguous
-- **platform**: Identified platform (web/android/ios/flutter/electron), or `null`
-- **query**: The user's original question / keywords
-- **intent**: one of the values listed below.
-
-> **Authoritative enum**: this file is the canonical source for the set of `intent` values search accepts. Callers (`docs`, `onboarding`) MUST only pass a value from this list and MUST map their own domain intent (e.g. docs' `slice-lookup`, onboarding's Path B free-text symptom) to one of these four. When this list changes, callers update their mapping tables. The list:
-
-- `error-code` — from `docs`: query contains a numeric error code. Prioritize `exact`; on miss, run full-text digit search before other strategies.
-- `pattern` — from `docs`: user asks for the official pattern of X, or compares X vs Y. Prioritize `tag`.
-- `feature` — from `docs` (implementation method lookup) or `onboarding` (user described a feature in natural language): need to find the corresponding slice ID. Prioritize `tag` / `product-keyword`. When multiple slices match, return summaries (`content_loaded: summary`) for the caller to present as a selection list.
-- `troubleshoot` — from `onboarding` Path B: user described a symptom in free text, need to find the slice(s) with relevant troubleshooting content. Match against slice troubleshooting sections and in-file `error_codes` sections.
-
----
-
-## Search Workflow
-
-### Step 1: Read the index
-
-Read `${CLAUDE_PLUGIN_ROOT}/knowledge-base/index.yaml` to get the catalog. Fields present at the index level:
-
-- **products**: `id`, `name`, `description`, `llms_file` (`llms_file` is used by the `docs` skill, not by search)
-- **cross_product_relations**: `id`, `products`, `slices`, `description`
-- **slices**: `id`, `name`, `tags`, `platforms`, `file`, `description`, `status`
-- **scenarios**: `id`, `name`, `slices`, `file`, `description`, `status`
-
-> **Fields NOT in index.yaml** (common misconception — don't assume them on the index):
-> - `error_codes` — lives inside the slice file body as an `## 错误码` / `## error_codes` section. To check it, the `file` path must be read.
-> - `related` — same: lives inside slice file frontmatter/body, not in index.
-> - `platform_files[platform]` — there is no such index field. Platform-specific content lives at `slices/{product}/{platform}/{ability}.md` on disk. To check availability, try reading that path; if it doesn't exist, there is no platform-specific slice for that pairing.
-
-> **Out of scope**: do NOT read llms files, do NOT fetch official doc URLs. If the caller needs a doc URL (e.g., fallback when no slice matches an error code), return `status: no_match` and let the calling skill (`docs`) do the llms.txt lookup.
-
-### Step 2: Five-Strategy Matching (by priority)
-
-Try strategies in order; collect matches. Earlier strategies have higher priority. Higher priority wins on rank; ties broken by tag-intersection count.
-
-| # | match_strategy | Triggers on | How to match |
-|---|---------------|-------------|--------------|
-| 1 | `exact` | query contains a numeric error code, a known slice_id like `chat/multi-instance`, or a scenario_id | (a) If an error code: look inside each slice's `file` for its `## 错误码` / `## error_codes` section and match the code. On `intent=error-code` miss, additionally grep all slice file bodies for the digit sequence (covers codes mentioned in troubleshooting prose). (b) If a slice_id or scenario_id: direct lookup in index arrays. |
-| 2 | `tag` | after exact (or alongside when no error code) | Extract keywords from the query (LLM does Chinese↔English bidirectionally on its own; consult the "Keyword Hints" table below only for non-intuitive SDK naming). Intersect with each slice's `tags`. Rank by intersection count. |
-| 3 | `product-keyword` | `product` is set AND tag didn't produce a high-confidence hit | Within that product's slices, fuzzy-match against `name` (higher weight) and `description`. Both Chinese and English terms. |
-| 4 | `cross-related` | `tag` or `product-keyword` produced a partial hit AND either (a) query contains signals of a second product, or (b) the initial hit set feels incomplete | Expand via `cross_product_relations` entries whose `products` overlap, OR follow the already-matched slice's `related` list (read from slice file) for one hop. Fold those into matches at medium confidence. |
-| 5 | `fuzzy` | no strategy 1-4 produced anything | Last-resort: use Keyword Hints for bidirectional expansion, match against scenario `name` + `description`, or match slice `description` without tag overlap. All hits are `confidence: low`. |
-
-#### Intent-driven priority adjustments
-
-| intent | Priority hint |
-|--------|---------------|
-| `error-code` | `exact` runs first; on miss AND `intent=error-code`, run full-text digit grep before strategy 2. Return immediately on strategy-1 hit. |
-| `pattern` | Run all strategies, but rank `tag` hits above `product-keyword`. |
-| `feature` | `tag` and `product-keyword` prioritized. If ≥ 4 matches total, downgrade all `content_loaded` to `summary` so the caller can present a selection. |
-| `troubleshoot` | Prefer slices whose body has a troubleshooting section relevant to the described symptom. Check in-file `error_codes` and 排障 sections when reading for rank. |
-
-#### product=null degradation
-
-When `product` is `null`:
-
-- Strategies `exact` and `tag` run normally (they don't need product). If either yields a `confidence: high` hit, return `status: matched` — do NOT trigger ambiguity.
-- `product-keyword` is skipped (it requires product).
-- `cross-related` and `fuzzy` run as full-catalog searches, capped at **5** matches, ranked by tag-intersection count (ties: `name` match > `description` match).
-- If all remaining results are `confidence: low` (fuzzy only) AND they span ≥ 2 products, return `status: ambiguous_product` with `ambiguous_candidates` listing the distinct products seen. DO NOT return low-confidence results — force the caller to disambiguate first.
-- If fuzzy hits stay within one product, return them normally as `matched`.
-
-### Step 3: Platform Filtering
-
-If `platform` is specified:
-
-- Filter results to slices whose `platforms` array contains the requested platform.
-- Keep cross-platform overviews (slices present at `slices/{product}/{ability}.md` with no platform subpath) as secondary results — they're generally useful.
-- Framework-to-platform mapping (applied at the caller's layer but repeated here as reference):
-  - React / Vue → Web
-  - Kotlin → Android
-  - Swift / Objective-C → iOS
-  - Dart → Flutter
-
-If the user needs code but hasn't specified a platform, surface that fact via `reason` (`"platform required for code example"`) — the caller asks.
-
-### Step 4: Progressive Content Loading
-
-Set `content_loaded` per match based on result-set size:
-
-- **Top 1-3 matches in `matches[]`** → read the full slice file at `${CLAUDE_PLUGIN_ROOT}/${CLAUDE_PLUGIN_ROOT}/knowledge-base/{slice.file}`; if a platform-specific file exists at `${CLAUDE_PLUGIN_ROOT}/knowledge-base/slices/{product}/{platform}/{ability}.md`, read that too. Set `content_loaded: full`.
-- **4+ matches** (typical on `intent=feature` ambiguity) → do NOT read each file; return index-level fields only (name, description, tags). Set `content_loaded: summary`. Let the caller present a selection.
-- **Slice with `status: planned`** → do NOT try to open its `file` (it likely doesn't exist yet). Use the index description. Set `content_loaded: index-only` and set top-level `status: status_planned` on the response.
-
-Record every file actually read in `matches[].file_paths_read`. This is the caller's canonical source for "which files grounded the answer".
-
-### Step 5: Build the response
-
-Fill the Response Contract based on Step 2-4 results:
-
-| Situation | response.status | matches | extras |
-|-----------|----------------|---------|--------|
-| Strategies 1-5 produced content-loadable hits | `matched` | populated; confidence derived from match_strategy | — |
-| Hits all pointed at planned slices | `status_planned` | populated with `content_loaded: index-only` | `reason: "slice <id> is status: planned"` |
-| product exists in index but has zero slices | `no_slice` | empty | `reason: "product <id> has no slices in KB"` |
-| product=null AND only fuzzy hits spanning ≥ 2 products | `ambiguous_product` | empty | `ambiguous_candidates: [...]`, `reason: "..."` |
-| Strategies exhausted, nothing relevant at any confidence | `no_match` | empty | `reason: "..."` (be specific — error code not found / no tag overlap / etc.) |
-
-> **Display rules live with the caller**, not with search. search only reports `file_paths_read`; it does not dictate how the caller quotes code, when to include snippets, or how to render confidence. See `../trtc-docs/SKILL.md` Step 3 and `onboarding/reference/path-a2-integrate.md` slice-loading block for those rules.
-
----
-
-## Keyword Hints (non-intuitive mappings only)
-
-Bidirectional Chinese↔English translation for common vocabulary is delegated to the LLM — it already understands that "弹幕" ≈ barrage/danmu, "美颜" ≈ beauty, "进房" ≈ enter-room. This table lists **only mappings that are NOT obvious from translation alone** — cases where the user's colloquial term differs from the SDK's internal naming.
-
-| User phrasing | SDK tag / keyword | Why it needs a hint |
-|---------------|--------------------|---------------------|
-| 互踢 | `kick-offline`, `multi-instance` | Not an SDK term; the root capability is multi-instance login conflict |
-| PK | `battle`, `BattleStore` | "PK" is colloquial; SDK names it battle |
-| 黑屏 | `setLiveID`, `black-screen` | The usual root cause is a missed `setLiveID` call, not obvious from "black screen" |
-| 跨房连线 | `co-host`, `CoHostStore` | SDK-proprietary term |
-| 连麦 | `co-guest`, `seat`, `CoGuestStore` | SDK uses `co-guest`, not the more literal `co-mic` |
-| 踢人 | `kick`, `kickUserOutOfRoom` | Exact method name is non-intuitive |
-| 音效 | `audio-effect`, `changer`, `reverb`, `ear-monitor` | Multiple sub-capabilities under one colloquial term |
-| 观众 | `audience`, `LiveAudienceStore` | Associated Store name is not what a user would guess |
-| 100211 / take-seat | `room-lifecycle`, standard room | Non-seat room internal sync; usually ignorable |
-| 100006 / live room only | `room-lifecycle`, `official-roomkit-api` | Live API called in standard conference room |
-| MediaPermissionCheck / getUserMedia 门禁 | `prejoin-check`, `device-control` | Lightweight browser gate vs Atomicx full prejoin test |
-| DeviceDetector / 两套设备检测 | `prejoin-check`, `device-control` | Dual detection paths need layered responsibilities |
-| 检查遗漏 / 业务流程 / 对照官方 | `conference/integration-audit` (meta slice) | Route to Path B intent F, not code review |
-| 进房失败 | `login-auth`, `room-lifecycle` | Conference Web join/create flow |
-| UserSig 过期 / 6206 | `login-auth` | IM auth layer; cross-product |
-
-For any phrasing NOT listed: let the LLM do its own Chinese↔English bridging, then match against `tags` / `name` / `description` directly. Do NOT expand this table defensively with obvious translations — it only pays for its maintenance cost on the non-obvious ones.
-
----
-
-## Edge Cases
-
-### Missing content
-
-| Scenario | status | Handling |
-|----------|--------|----------|
-| Product exists in index but has no slices (e.g., call, conference at early phase) | `no_slice` | `matches: []`, `reason: "product <id> has no slices in KB"` — caller (docs) falls back to llms.txt |
-| Matched slice has `status: planned` in index | `status_planned` | `matches[].content_loaded: index-only`, use index description; caller decides fallback |
-| Platform requested but no platform-specific file at `slices/{product}/{platform}/{ability}.md` | `matched` | Return product-level overview with `content_loaded: full`, `reason: "no platform-specific file for <platform>"` — caller tells the user in their own language that the platform-specific code example isn't yet in the KB; do NOT synthesize code |
-
-### Cross-product questions
-
-| Scenario | Handling |
-|----------|---------|
-| "直播+弹幕" (Live+Chat) | `tag` hits live/* + query contains chat signal → `cross-related` expands via `cross_product_relations` → matches include slices from both products |
-| "视频通话+聊天" (Call+Chat) | Same pattern: `tag`/`product-keyword` initial hits + `cross-related` expansion |
-| "该用哪个产品？" | Out of scope — return `no_match` with `reason: "product-selection question, not a slice lookup"`; caller (root / onboarding) handles product identification |
-
-### Search ambiguity
-
-| Scenario | Handling |
-|----------|---------|
-| Vague single-term query like "消息" with product=null | `tag` returns many chat/* hits → `matched` with `content_loaded: summary` for the top 3-5; caller presents selection |
-| Mixed Chinese-English "怎么 kick offline" | LLM bridges "kick offline" → Keyword Hints row for 互踢 → `tag` match |
-| Only an error code like "6208" | `exact` (strategy 1) → `matched`, confidence `high` |
-| Error code not in any slice | `no_match`, `reason: "error code <code> not in any slice.error_codes or body text"` — caller (docs) falls back to llms.txt |
-
-### Platform edge cases
-
-| Scenario | Handling |
-|----------|---------|
-| Conceptual question, no platform needed | Return product-level overview only; `platform` argument can be null |
-| Needs code but no platform specified | Return `matched` with whatever was found; set `reason: "platform required for code example"` — caller asks |
-| React/Vue → Web, Kotlin → Android, Swift/ObjC → iOS, Dart → Flutter | Caller normalizes; search accepts the normalized platform |