Skill Sublation
技能扬弃 Sublation V5.0 完整版。把执行经验沉淀为观测、候选、验证、多 Agent 协作面板、默认关闭的受控唤醒意图、Skill 分类与 shadow 精准建议、用户批准、晋升和观察窗的可审计治理链路;仅在用户明确说出 sublation 或扬弃技能时触发。
Install
npx skills add Sven-Mirana/sublation --skill skill-sublationgit clone --depth 1 https://github.com/Sven-Mirana/sublation.git /tmp/sublation
mkdir -p ~/.claude/skills
cp -r /tmp/sublation/. ~/.claude/skills/skill-sublationPersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# Skill Sublation v5.0 Sublation 是技能治理宪法,不是自动改技能的捷径。 V5.0 在既有治理链上增加三项能力:项目本地、可限缩或拓展的多 Agent 协作面板;默认关闭且带受控关窗的 Agent 自动唤醒意图层;晋升后随 Skill 携带、由 central router 统一解释的分类画像与精准建议 shadow。三项能力都不改变权限边界:面板席位不等于治理复核授权,唤醒意图或 claim 不等于 Agent 已运行或已回复,路由建议不等于宿主选择,更不等于 Skill 已被调用。V5 候选不安装真实 host adapter、launchd 或 cron;没有另行集成和授权时始终保持关闭。 它处理的核心矛盾是:Agent 会在执行中发现技能缺陷、边界裂缝和可吸收经验,但正式技能不能被即时、静默、无证据地改写。因此所有改进必须先进入候选层,用证据证明价值,再由用户守住生产门。 标准链路: ```text Observation -> Candidate -> Validation -> Review-seat reports -> Coordinator unified brief -> User decision -> Promotion -> Observation window -> Closure ``` ## 1. 宪法 ### 1.1 三条根本原则 1. **所有 skill 开发都必须走 sublation 全链路** 创建、修改、吸收、合并、拆分、删除、发布前清理,都必须留下观测、候选、审计、复核、用户决策、回滚和观察窗证据。小改动可以批处理,但不能绕过账本。 2. **sublation 必须自我扬弃** sublation 不是只管别人的治理工具。它自身也是被治理的对象。每次流程暴露裂缝,都要回流为框架改进,并记录在 `references/sublation-self-evolution.md`。 3. **不是管别人,是先被管** 参与治理的 agent 在要求其他 skill 接受治理前,先接受同一套约束。默认本地席位是 Hermes、Codex、Claude Code;其他部署可以换成自己的 agent,但 builder、independent verifier、reviewer 默认必须是三个不同 actor。只有用户对当前 run 显式授权,才允许单代理模式,并且必须在 `review_policy` 中披露证据密度下降。自报“已完成”不算证据;grep、read-back、diff、hash、audit、fixture 才算证据。 ### 1.2 扬弃的判定 扬弃不是“又多一个候选”,而是旧能力被保留,同时出现可证明的正向增量: - 能力更强或覆盖更广; - 边界更清; - 稳定性、安全性或可维护性更高; - 治理质量提升; - 旧 workflow 有 fallback 或明确的用户批准。 没有正向增量,最多记录 observation;退化不叫扬弃。 ### 1.3 经验不是权威 任何 agent、外部评估器、benchmark、scorecard 都只能提供证据,不能替代用户决策。晋升权只属于用户,除非用户在当前任务中明确委托 Agent 执行已批准的低风险合入。 ## 2. 硬边界 ### 2.1 正式技能默认只读
Details
- Skill name
skill-sublation- Repository
- Sven-Mirana/sublation
- License
- MIT
- Repo updated
- 26 August 2026
FAQ
What does the Skill Sublation skill do?
技能扬弃 Sublation V5.0 完整版。把执行经验沉淀为观测、候选、验证、多 Agent 协作面板、默认关闭的受控唤醒意图、Skill 分类与 shadow 精准建议、用户批准、晋升和观察窗的可审计治理链路;仅在用户明确说出 sublation 或扬弃技能时触发。
How do I install the Skill Sublation skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add Sven-Mirana/sublation --skill skill-sublation. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Skill Sublation popular?
It lives in Sven-Mirana/sublation, which has 279 GitHub stars and 2 forks.
Related skills
Agentii Investment Intelligence
210agentii-ai/agentii-investment-intelligence
Institutional-grade equity research skills for AI agents. 80 Claude-type skills across 14 verticals (equity-research-core, models-and-pitches, bio-pharm, scenarios, quantitative-analysis, idea-generation, options-derivatives, business-intelligence, industry-analysis, macro-strategy, portfolio-strategy, technical-analysis, risk-and-psychology, trading-as-business) powered by agentii.ai's agent-use-ready SEC filing data plane — SEC filings and 15K+ earnings call transcripts, 15.99M XBRL facts, and company profiles for 1,146+ US-public-equity tickers. Features the three-layer retrieval protocol (
Mcd Party Officer
244yzg1226/mcd-party-officer
麦当劳门店派对/包场/拼团的选址、成团调度与筹备助手。当用户想办生日派对、亲子派对、主题派对、品鉴会、读书会、麦麦体验营,或说「麦当劳派对」「派对拼团」「包场」「生日会怎么办」「哪家店能办派对」「还差几个人成团」「包场还是拼团划算」「帮我写拼团招募文案」「做一张派对邀请函」「孩子派对吃什么不超标」时使用。基于麦当劳官方 MCP(mcp.mcd.cn)的 party 系列工具:扫描全城场次、估算成团概率、对比包场与拼团成本、按官方营销活动日历挑最佳日期、审查儿童餐营养、生成招募卡与邀请函。
Skills Constitution
234jiabaobei/skills-constitution
当 Agent 接到专业任务(编码/爬虫/文件操作/API调用/数据分析/文档/部署/推送等)时,强制先查记忆层和技能索引,有匹配必用、无匹配必搜、答复时自动推荐(排除已装)。用于防止 Agent 跳过技能直接硬扛通用能力。跨平台通用(WorkBuddy/Claude/ChatGPT/Cursor/Gemini 等 20+ 框架)。完整版本史见 CHANGELOG.md。
LLM Wiki
123praneybehl/llm-wiki-plugin
Build and maintain an LLM-curated knowledge base from papers, articles, transcripts, notes and project findings. Ingest sources into linked Markdown, answer questions with citations, lint or upgrade a wiki, and preserve useful synthesis across sessions and agents. Also capture verified task experience, consolidate success/failure patterns, and propose, evaluate, apply or roll back evidence-linked skill improvements. Trigger on "add this to my wiki", "what does the wiki say", "learn from this task", "improve this procedure", /wiki:learn or /wiki:evolve. Works on a local folder or a hosted wiki
Pitchcraft
201moshuying/pitchcraft
Structured persuasion for tech leads, PMs, and founders—not activity logs. Five scenarios (kickoff, status update, wrap-up, investor pitch, solution selling) on one 5-part framework (Hook→Context→Proposal→Evidence→Ask). AI prompts for missing materials and audience context; pre-submit checklist. Claude Code plugin; Cursor, Codex, and chat via prompts.
Ancient Rap Battle
19110291124jcjc-dotcom/ancient-rap-battle
古人说唱 Battle:把两个历史上的死对头(诸葛亮 vs 司马懿、岳飞 vs 秦桧……)做成一条 AI 说唱对决短视频,从选题写词(每句配史料注释)、Suno 做歌、Midjourney 定妆、即梦 Seedance 2.5 网页版分段生成(提示词按半句精确到 0.01 秒写歌词,口型一次对好),到自动逐镜头对齐、拼接、字幕包装、封面和发布文案一条龙。用户说"做一期 A vs B"、"古人说唱 battle"、"历史人物说唱"、"rap battle 视频",或者要改已有一期的歌词、镜头、字幕、封面时使用。Make an AI rap-battle short between two historical rivals, end to end, with Seedance 2.5 (or MiniMax H3).
Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.