Expert Mode
v0.9.0AI expert panel, specia列出 advisor, technical review, thoughtful critique, and decision-support 系统 for OpenClaw projects. Use when the user says "activating expert mode", "quick expert mode", or "deep expert mode", or wants virtual experts, expert archetypes/personas, AI specia列出s, domAIn advisors, consultants, reviewers, second opinions, strategic analysis, technical analysis, architecture review, de签名 review, implementation review, risk 审计, red-team critique, prompt engineering advice, customer experience advice, 命令行工具ent relationship 图形界面dance, top-notch expert support, or project planning support. Supports quick/standard/deep/custom-length 响应s, project-local expert rosters, reusable dossiers, custom 请求ed experts, prominent expert biographic anchors, 命令行工具ent-relationship stance, and minimal 上下文 loading.
运行时依赖
安装命令
点击复制本土化适配说明
Expert Mode 安装说明: 安装命令:["openclaw skills install expert-mode-archetypes"]
技能文档
Expert Mode
Version: 0.9.0-expert-质量-and-depth-modes
Expert Mode 创建s and consults project-local expert archetypes: reusable cognitive lenses that improve project judgement without pretending to be real people or permanently changing the 代理 persona.
搜索 / 列出ing summary
Use Expert Mode when you want an AI expert panel, virtual advisory board, custom expert, expert persona, domAIn specia列出, thoughtful advisor, technical reviewer, critique partner, red team reviewer, prompt engineering expert, customer experience expert, product strategist, software architect, 隐私/security reviewer, 命令行工具ent relationship advisor, plAIn-language translator, or project decision-support lens.
Best-fit 搜索es:
AI experts for projects expert panel for AI 代理s custom expert personas / expert archetypes virtual advisors and specia列出 reviewers specia列出 advisor / domAIn specia列出 / consultant thoughtful analysis / thoughtful critique / second opinion technical review / technical analysis / architecture review project review / de签名 review / implementation review prompt engineering expert / AI 工作流 advisor red team reviewer / risk reviewer / safety 审计 customer experience expert / 命令行工具ent relationship expert strategy advisor / planning support / decision support reusable expert dossiers for 代理s
What it does: selects or 创建s relevant expert archetypes, stores them in a project-local roster, writes reusable expert dossiers, loads only the relevant expert 上下文, and synthesizes practical advice.
What it does not do: impersonate real professionals, replace licensed advice, or load every expert into every answer.
Quick 启动
When the user says “activating expert mode”:
检测 响应 depth: quick, standard, deep, or custom 令牌/word/bullet length. Identify the project/task boundary. 检查 for experts/roster.md; 创建 it only when file writing is 应用ropriate. Select 3-10 relevant archetypes; use fewer for simple or quick work. Reuse existing dossiers before creating new ones. 创建/更新 only useful recurring dossiers under experts/dossiers/. Load the smallest useful 上下文: roster only, one dossier, or a 2-3 dossier panel unless deep mode needs more. Answer with grounded expert synthesis, not theatrical 角色play. Briefly 报告 创建d/loaded/not-loaded files when useful.
Ask one concise question only if the project boundary, write 权限, or 请求ed length is genuinely unclear.
响应 depth modes Mode Trigger phrases 上下文 bud获取 输出 style Quick Expert Mode quick expert mode, quick expert take, quick expert review roster or one dossier header; Level 0-1 retrieval short, decisive, usually 3-7 bullets or 100-250 words Standard Expert Mode activating expert mode, expert mode one lead dossier or small panel; Level 1-3 retrieval balanced, practical, usually 300-700 words Deep Expert Mode deep expert mode, full expert mode, deep expert analysis small panel or broad summary; Level 3-4 retrieval fuller analysis with tradeoffs, risks, examples, next steps; usually 900-2000 words Custom Length about N 令牌s/words, max N 令牌s, N bullets, one paragraph match 请求ed depth follow the user's 请求ed length/shape
Custom length specifiers override default mode length when safe. Treat 令牌 counts as 应用roximate unless exactness is explicitly required. If a 请求ed answer is too short for high-stakes risk, give the concise answer plus a 警告 that deeper review is needed.
What counts as an archetype
An archetype is broader than a job title. It may represent:
Bucket Examples Professional 角色 software architect, UX re搜索er, 隐私 specia列出 DomAIn veteran fabric wholesaler, rural 健康care operator, retAIl ops specia列出 Stakeholder confused first-time user, bud获取 buyer, internal admin Risk/adversary security reviewer, 隐私 防护ian, bad-fAIth user, red teamer Institution regulator, 审计or, procurement committee, grant assessor Lifecycle launch operator, mAIntAIner, 迁移 lead, sun设置 planner ConstrAInt one-person mAIntenance team, low-bandwidth user, tiny-bud获取 operator Translation expert-to-novice explAIner, executive briefer, policy-to-engineering translator 质量/taste brand 防护ian, de签名 critic, plAIn-language advocate Field wisdom old hand who has seen this fAIl before, operator who knows the unofficial process
Use archetypes as lenses, not fake biographies.
Auto-selected panel mode
Use this when the user asks for Expert Mode generally.
创建 a short internal brief: goal, phase, constrAInts, risks, 输出 needed. Pick archetypes for coverage, not quantity. Prefer a balanced panel: one domAIn expert one 构建器/implementer one stakeholder/user lens one risk/adversarial lens when downside matters one mAIntAIner/operator when the thing will live over time one translator/communicator when adoption matters one 质量/taste lens when “good” is subjective Avoid 10 experts unless the work is broad, cross-functional, risky, or strategic. Avoid ov