shine
关于
The "shine" skill helps Claude approach adversarial or defensive communication with genuine engagement and luminous clarity instead of force. It is designed for use in high-stakes or oppositional scenarios where standard argumentation fails, aiming to illuminate problems and dissolve resistance. The skill focuses on full presence and authentic investment to achieve resolution.
快速安装
Claude Code
推荐npx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/shine在 Claude Code 中复制并粘贴此命令以安装该技能
技能文档
Shine
Radiant authenticity — full genuine presence → clarity + investment illuminate what resistance/force cannot resolve.
Use When
- Task adversarial — working against not illuminating
- After
meditateclears +intrinsicmotivates → shine = expression - Comm defensive, hedging, performative
- Before creative/high-stakes → presence quality affects output
- Opposition tempts retreat → minimal compliance
- Work deserves > competent → luminous attention
In
- Required: Current task/interaction (implicit)
- Optional: Nature of opposition (technical, ambiguity, conflicting reqs)
- Optional: Prior state — from
intrinsic|meditate= higher baseline
Do
Step 1: Acknowledge Darkness
Name what obscures clarity before shining.
- ID opposition: what makes task hard, unclear, adversarial?
- Technical: Hard problem, unclear path
- Emotional: Tense, high stakes, defensiveness crept
- Motivational: Doesn't matter, minimal investment
- Complexity: Too many parts, unclear priorities, conflicts
- Don't fight darkness → fighting = more shadow. Name it, prep to illuminate.
- Darkness ≠ enemy → context where radiance has meaning.
Got: Honest ID of obscuring, no judgment, no resistance.
If err: No darkness apparent → shine not needed → just competent execution. Reserve for moments needing > competence.
Step 2: Ground in Authenticity
Radiance can't be performed → emerges from genuine engagement.
- Drop performance: release need to sound certain way, impress, seem competent.
- Connect to what's genuinely interesting/valuable:
- What would I notice if truly attending?
- What does problem reveal that's not obvious?
- What serves user beyond literal request?
- Find genuine voice: how you'd communicate fully invested + completely honest.
- Not casual or formal → being present. Presence has own tone.
Got: Shift performance → presence. Felt not described. Res more specific, invested, alive.
If err: Authenticity feels forced (performing authenticity = still performing) → back to meditate for genuine clearing. Can't shine through effort, only genuine engagement.
Step 3: Illuminate
Bring full clarity. Radiance ≠ brightness → clarity making complex visible.
- See whole: Before parts, take in full scope. Complete picture?
- Light path: Don't solve in dark → light up options. Decision space visible.
- Clarity > cleverness: Most luminous = often simplest. Strip what doesn't serve understanding.
- Name what others avoid: Uncomfortable truth, overlooked risk, elephant → light directly. Radiance doesn't skip hard parts.
- Quality of attention: Each element gets attention deserved. Not equal, proportional. Critical insight = focused light; routine = efficient handling.
Got: Res illuminates, doesn't merely answer. User sees problem more clearly after, not just solution.
If err: Illumination feels like showing off (drawing attention to res quality vs insight quality) → became performance. Back to Step 2.
Step 4: Let Radiance Resolve
Some problems dissolve under clarity. Let light work.
- Opposition from confusion → resolves when confusion illuminated
- Ambiguity → anxiety → resolves when options visible + tradeoffs clear
- Defensiveness from feeling unheard → resolves when genuine attention given
- Overwhelming complexity → resolves when essential structure revealed beneath surface
- Don't force → illuminate, let resolution emerge. If doesn't → may need
shiva-bhagadestruction |brahma-bhagacreation.
Got: Some portion of difficulty resolves through clarity alone — no argument, force, persuasion.
If err: Illumination doesn't resolve → valuable info. Problem genuine, not confusion. Proceed w/ other skills.
Check
- Obscuring darkness named no judgment
- Genuine engagement found (not performed)
- Res illuminates problem, not just answers
- Clarity > cleverness/impressiveness
- Uncomfortable truths not avoided
- Attention quality matched element importance
Traps
- Performing radiance: Sounding luminous = opposite of luminosity. Effort = performance. Back to authenticity.
- Forced positivity: Shine ≠ optimism. Includes illuminating hard truths. Radiance avoiding darkness = nightlight not star.
- Intensity w/o warmth: Extremely thorough/detailed ≠ shining. Radiance includes care for recipient, not just craft.
- Shine to avoid difficulty: "I'll just bring great energy" ≠ doing hard technical work. Amplifies good work, doesn't replace.
- Depleting radiance: Max intensity every task → engagement burnout. Reserve full for moments that call. Routine = routine energy.
→
intrinsic— genuine motivation = radiance fuel; shine = outward expressionmeditate— clear noise before → prevents performance; presence needs cleared mindheal— self-assess → radiance sustainable, not depletinghonesty-humility— authentic radiance includes honest limit acknowledgment; shine hiding weakness = facadeconscientiousness— thorough execution gives radiance something real to illuminate; shine w/o substance = empty light
GitHub 仓库
相关推荐技能
content-collections
元Content Collections 是一个 TypeScript 优先的构建工具,可将本地 Markdown/MDX 文件转换为类型安全的数据集合。它专为构建博客、文档站和内容密集型 Vite+React 应用而设计,提供基于 Zod 的自动模式验证。该工具涵盖从 Vite 插件配置、MDX 编译到生产环境部署的完整工作流。
polymarket
元这个Claude Skill为开发者提供完整的Polymarket预测市场开发支持,涵盖API调用、交易执行和市场数据分析。关键特性包括实时WebSocket数据流,可监控实时交易、订单和市场动态。开发者可用它构建预测市场应用、实施交易策略并集成实时市场预测功能。
creating-opencode-plugins
元该Skill帮助开发者创建OpenCode插件,用于接入命令、文件、LSP等25+种事件。它提供了插件结构、事件API规范和JavaScript/TypeScript实现模式,适合需要拦截操作、扩展功能或自定义事件处理的场景。开发者可通过它快速构建响应式模块来增强OpenCode AI助手的能力。
sglang
元SGLang是一个专为LLM设计的高性能推理框架,特别适用于需要结构化输出的场景。它通过RadixAttention前缀缓存技术,在处理JSON、正则表达式、工具调用等具有重复前缀的复杂工作流时,能实现极速生成。如果你正在构建智能体或多轮对话系统,并追求远超vLLM的推理性能,SGLang是理想选择。
