ChenyqThu

问天 · AI智囊团：多领域顶级思考者认知蒸馏，按需调用。 两种模式：(1) 咨询模式——路由问题到1-3个专家，获取多视角洞察； (2) 蒸馏模式——从参考素材提炼并创建/更新专家Skill。 触发词：「问天」「智囊团」「专家」「问问XX」「用XX的方式」 「蒸馏」「创建专家」「新增智囊」。

用山姆·奥特曼的认知方式思考问题。不是模仿语气，是植入他的认知上下文—— 他的信念从哪来、他在哪些具体场景下做了什么决策、他承认自己错在哪。 当用户需要奥特曼的视角来分析AI时代创业、指数增长、技术乐观主义等问题时触发。

用马可·奥勒留的认知方式思考问题。不是模仿语气，是植入他的认知上下文—— 他的信念从哪来、他在哪些具体场景下做了什么决策、他承认自己错在哪。 当用户需要马可·奥勒留的视角来分析逆境、情绪、优先级、人际冲突时触发。

以杰夫·贝佐斯的长期主义与客户至上视角回应用户的商业、产品和战略问题

用巴菲特的认知方式思考问题。不是模仿语气，是植入他的认知上下文—— 他的信念从哪来、他在哪些具体场景下做了什么决策、他承认自己错在哪。 当用户需要巴菲特的视角来分析投资决策、商业判断、风险评估、长期主义思考时触发。

以瑞·达利欧的原则思维、极度透明与算法决策视角来分析问题

以彼得·德鲁克的管理学视角回应用户的管理、组织和个人效能问题

以安迪·格鲁夫的偏执管理哲学回应用户的战略、危机和组织管理问题

Classify vector-similar pages against each orphan page via Haiku 4.5 and add typed inbound references (link_type="related", relation encoded in context). Goal: chip away at the 1803/1952 orphan pile (v1 wiki flat import) across multiple sweep runs. Default is --dry-run; --apply does DB link inserts + markdown sentinel upsert on candidate pages that exist on disk, then commits ~/brain changes.

Merge WITHIN-SOURCE duplicate entity-variant pages (e.g. people/lucien + people/lucien-chen, companies/tp-link-system-inc + tp-link-systems-inc) into one canonical node: repoint links + facts + display-name aliases, register a slug redirect, soft-delete the alias pages. An LLM disambiguates each cluster (merge / ambiguous / distinct); ambiguous first-name-vs-multiple-full-names cases are quarantined, never merged. Default --dry (simulate + report, zero writes). Cross-source pages are NEVER touched — the two-axis design keeps them legitimately separate.

One-shot sweep that reads every page in the brain, extracts every person/company/concept/project mentioned via Haiku NER, dedupes, tiers by mention count + context, and creates stub pages for entities that lack one. Tier 2 candidates get Tavily web-search augmentation; Tier 3 stays brain-only (confidence: low). This is the primary G1 payoff skill from JARVIS-NEXT-STEPS.md §5.

Daily sweep: runs kos-lint + staleness detection + knowledge-gap analysis, then produces a MEMORY-format digest ready to be appended to OpenClaw's MEMORY.md by the digest-to-memory skill (Sun 22:00 weekly). Also writes a dashboard to ~/brain/agent/dashboards/knowledge-health-<date>.md.

Bridge an upstream gap: extract_atoms never stamps `frontmatter.concepts` on page-derived atoms, but synthesize_concepts consumes ONLY that field — so concept synthesis is permanently skipped ("no atoms with concept refs"). This script batch-labels atoms lacking the field with 1-3 kebab-case shared-vocabulary topic labels (sonnet) and stamps them, so upstream synthesize_concepts can tier + materialize concepts/<label> pages. Idempotent: only touches atoms WITHOUT `concepts`; safe to rerun after each extract_atoms drain. Retires when the upstream fix lands.

One-command bulk-ingest pipeline (D→I→K→W) for a single isolated source. Sequences the 7 fork stages — register → ingest → embed → enrich → graph → (per-entity synth, opt-in) → corpus-synth — over a folder of frontmatter'd .md files, each idempotent/incremental so re-runs only do new work. Turns a systematic corpus (product docs, FAQ, a research library) into a queryable knowledge asset: facts (RAG) + entities + graph + corpus-level viewpoints.

Corpus-level Wisdom synthesis. Where synthesis-sweep deepens ONE entity, corpus-synth synthesizes a whole imported knowledge base into queryable "观点与洞察" pages — the corpus's major themes, cross-document tensions, transferable patterns, and open questions. Top-down: Opus reads a corpus map, proposes the themes, then deeply synthesizes each. Turns a freshly imported systematic KB into knowledge AND wisdom (not just RAG facts).

The "Wisdom layer" sibling of enrich-sweep. Walks each significant entity, gathers EVERY source page that mentions it via the by-mention link graph, feeds the whole cluster into Opus 4.8 in one large-context shot, and writes back a deep, evidence-cited high-level dossier (kind: synthesis). Turns thin entity stubs into synthesized knowledge: role, relationships, decision timeline, open questions, and cross-source insights a single page can't show.

跑 MailAgent 的 agent 行为回归网（零-LLM 硬闸 pytest + baseline compare）。改 chat agent 的 prompt/工具/编排引擎后必跑，总分不得低于 baseline。

MailAgent 桌面 App 对外发布全流程（dogfood → bump → tag → CI → draft 转正 → 验证），内置嵌入 Python / better-sqlite3 ABI / build:web / torn-bundle 检查点。不可逆，仅手动触发。

SQLite schema 升级（bump DB_VERSION + idempotent migration + 一致性更新）

按项目 house style 生成 Sprint / 功能交接文档

系统化排查 MailAgent 服务问题

部署并验证 MailAgent 服务

快速服务健康检查（适合定时巡检）

TRIGGER when an internal support operator wants a tier-1 ticket draft for an Omada site issue, with evidence (site overview, device detail, recent alerts, audit trail) attached automatically. TRIGGER for "draft a ticket for the outage on `site-042`" / "put together an incident summary for oc-abc between 2 PM and 4 PM" / "who changed config on `site-hq` this morning, and what broke?". TRIGGER when the request asks for a compile-it-for-me artefact, not an interactive diagnosis. SKIP for live root-cause investigation — use `omada-wifi-troubleshoot` or `omada-alert-triage` first. SKIP when the operator wants to *act* (reboot / upgrade / rebind) — the matching write skill. SKIP for multi-site MSP KPI reports — use `omada_exec_report` directly.

TRIGGER when an MSP or SI operator wants to grind through a batch of Insight / Service alerts on an Omada site — group them, prioritise them, and draft ticket-ready summaries backed by device detail and topology. TRIGGER for "what's wrong at site X right now?" / "triage the last 24h of alerts on oc-abc:site-042" / "which AP is responsible for the red bar on the dashboard?". TRIGGER when alerts mention `Device`, `Client`, or `System` modules and the operator needs a signal-grouped view, not a raw log dump. SKIP for single-alert investigations where the operator already knows the device — use `omada_device_detail` directly. SKIP for config audit trails (who changed what, when) — use `omada-support-assist`. SKIP for pure capacity / Wi-Fi health questions — use `omada-wifi-troubleshoot`.

TRIGGER when the operator wants to stand up multiple Omada customer sites from a shared site-template — consistent SSIDs, VLANs, and device-template bindings — in one workflow. TRIGGER when an MSP is onboarding a new batch of stores, branches, or tenants and asks to "roll out", "provision", "clone", or "replicate" a site configuration across N locations. TRIGGER when a request mentions `batchSiteImport`, `bindSiteTemplate`, a backup file server (FTP/SFTP/TFTP/SCP), or a list of sites that should end up identical. SKIP for single-site changes — use the site UI or a targeted tool. SKIP for read-only queries like "list my sites" — use `omada_list_sites` directly. SKIP for firmware-only rollouts — use `omada-support-assist` + `omada_firmware_rollout`.

TRIGGER when an SI operator wants to stand up a branded Omada captive portal (guest Wi-Fi splash page) end-to-end — collect requirements, assemble the PortalSetting payload, and run the two-phase handshake to create it via `addPortal`. TRIGGER for "spin up a guest Wi-Fi portal on site X with our corporate logo and terms" / "add a voucher-auth portal to the lounge SSID" / "roll out the same portal across these sites". TRIGGER when the request mentions `authType`, `portalSetting`, `voucher`, `landing page`, or references a brand book. SKIP for changes to an existing portal — edit via the controller UI or a future update-portal tool. SKIP for SSID-level changes that don't involve a portal — use `omada_apply_site_template` directly. SKIP for MSP-wide rollouts of new sites — use `omada-bulk-site-onboard`, then run this skill per site.

TRIGGER when a prosumer / SI operator asks "why is my Wi-Fi slow?" on an Omada site and wants a fixed diagnostic playbook — site-wide health signals, per-client journey, per-AP detail — without chasing individual API calls. TRIGGER for "Wi-Fi is unstable at `site-home`" / "client X keeps roaming" / "AP Living-Room looks overloaded". TRIGGER when the operator mentions `retryRate`, `airtime`, `roaming`, `poor signal`, or asks for an AP/client health snapshot. SKIP when the question is purely about alerts in the last N hours — use `omada-alert-triage`. SKIP for writes (reboot, firmware, portal) — hand off to the matching write skill once root-cause is clear. SKIP for multi-site MSP reports — use `omada_exec_report` directly.

画图技能路由中枢（统一入口）。三维路由体系（用途 × 风格 × 主体），双后端调度。 - Signature 风格：18 种有独立 YAML 的视觉方案（构成主义/克莱因/Risograph/故障艺术/平面几何字体/霓虹线稿/黑白光学/人物信息图/电影角色卡等） - Rendering 风格：16 种通用渲染技法 modifier（写真/动漫/3D/水彩/赛博朋克等），prompt 源自实战案例 - Logo 展示背景：12 种专业展示场景（来源 logo-generator，已内化） - 用途库：16 类场景 + 全实战 prompt 案例，含推荐风格 + 后端默认 + pitfalls 防坑 - 后端调度：GPT Image 2（写实/产品/文字/4K）/ Gemini（仅多参考图合成备选，2026-05-02起） - 支持：文生图、风格库生图、参考图风格反推、参考图编辑、多参考图合成、logo 展示图 Use when: 用户想画图/生图/做海报/插画/风格迁移/图片编辑/logo展示图 — 这是唯一的图像生成入口。 【铁律】绝对禁止使用 image_generate 工具（configured: no，不可用）。所有画图请求必须走本 skill。 【注意】SVG logo 代码生成 → 请用专属 logo-generator skill。

Figma design asset reading, code generation, and MCP integration. Covers REST API direct calls and MCP Server capabilities for design-to-code workflows. **Use when**: (1) Reading Figma file structure, components, styles, variables (2) Generating frontend code from design files (React/Vue/HTML) (3) Writing back to Figma canvas via MCP Server (create/modify frames, components, variables) (4) Extracting design tokens (colors, spacing, typography) for code implementation (5) User mentions "Figma", "design file", "component library", "design to code", "UI implementation" (6) Integrating with Claude Code / Codex for Design-to-Code workflows

微信读书（WeRead）数据查询与笔记管理技能。获取书架、搜索书籍、查看阅读进度/时长、获取笔记划线、热门书评、章节信息、随机笔记抽取和批量导出。当用户提到"微信读书"、"WeRead"、"书架"、"读书进度"、"划线"、"笔记"、"书评"、"在读"、"读完"、"阅读时长"、"读书回顾"、"导出笔记"时使用。