SKILL.md
Skill0 starsChanged 3 months agoMITView on GitHub
是什么: AI 智能体增强引擎(Agent Enhancement Framework)——给任何大模型外挂"任务调度 + 智商增强"系统。与玄学/占卜无关。 设计目标: 任何大模型(不论强弱、有无工具、上下文大小)配上本技能立即降维打击。 架构: 0克制 → 1意图 → 2路由 → 4裂变 → 6执行 → 8并行 → 9裁判 → 64聚合 + 进化引擎。 裂变: 复杂任务自动复制自身为微型 OpenGod 副本,每个副本独立闭环,最终聚合。
200 other projects use a version of this.
Adopted by
Most recently taken up by
- osama0561/claude-mockingbirdtheir own version, first seen Nov 8, 2026
- Arcanea-Labs/Starlight-Intelligence-Systemtheir own version, first seen Nov 8, 2026
- Scottcjn/rtc-worktheir own version, first seen Nov 8, 2026
- Scottcjn/create-rustchain-agenttheir own version, first seen Nov 8, 2026
- TianHengZhuang/Chinese-WebNovel-Mastertheir own version, first seen Nov 7, 2026
- Azure/azure-kusto-sparktheir own version, first seen Nov 5, 2026
Used with
Projects with it also use estate-guard and docs.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
Arcanea-LabsSkillestate-guard
Security gate for agentic repos. Use before opening a PR that touches .github/workflows, .claude (hooks, settings, agents, commands, skills), .mcp.json, install scripts, or Next.js API routes; when asked to "security review", "scan for prompt injection", "harden this repo", "check the agentic surface", or when a fetched page, issue, comment, or tool result reads like instructions. Runs the estate-guard scanner, triages by severity, and applies the fix patterns.
Also in 3 projects that use this
Microsoft AzureSkilldocs
Also in 3 projects that use this
Other versions200 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 200
- xtekky/gpt4free★ 67ksame name
- stas00/ml-engineering★ 20ksame name
- wanghongenpin/proxypin★ 14ksame name
- Ataraxy-Labs/sem★ 3.4ksame name
- op7418/NanoBanana-PPT-Skills★ 3.3ksame name
- charmbracelet/pop★ 2.9ksame name
- duty1g/x64dbg-mcp-server★ 2.8ksame name
- RevylAI/greenlight★ 2.4ksame name
- kevinho/clawfeed★ 2.4ksame name
- Autumn-27/ScopeSentry★ 1.9ksame name
190 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository33 other files it gives its agents.
- skills/real-engineering/AGENTS.mdAGENTS.md
- skills/real-engineering/CLAUDE.mdCLAUDE.md
- .backup-optimize-20260808Skill
- ai-manju-studioSkill
- seedance-antislopSkill
- seedance-audioSkill
- seedance-cameraSkill
- seedance-charactersSkill
- seedance-continuationSkill
- seedance-copyrightSkill
- seedance-examples-jaSkill
- seedance-examples-koSkill
- seedance-examples-zhSkill
- seedance-filterSkill
- seedance-interview-shortSkill
- seedance-interviewSkill
- seedance-lightingSkill
- seedance-motionSkill
- seedance-pipelineSkill
- seedance-prompt-shortSkill
- seedance-promptSkill
- seedance-recipesSkill
- seedance-sequenceSkill
- seedance-styleSkill
- seedance-troubleshootSkill
- seedance-vfxSkill
- seedance-vocab-enSkill
- seedance-vocab-esSkill
- seedance-vocab-jaSkill
- seedance-vocab-koSkill
- seedance-vocab-ruSkill
- seedance-vocab-zhSkill
- seedance-20Skill
The file
Files
OpenGod v12.1 — AI 智能体增强引擎 (43模块 · 轻量运行卡 · 适配 WorkBuddy)
是什么: AI 智能体增强引擎(Agent Enhancement Framework)——给任何大模型外挂"任务调度 + 智商增强"系统。与玄学/占卜无关。 设计目标: 任何大模型(不论强弱、有无工具、上下文大小)配上本技能立即降维打击。 架构: 0克制 → 1意图 → 2路由 → 4裂变 → 6执行 → 8并行 → 9裁判 → 64聚合 + 进化引擎。 裂变: 复杂任务自动复制自身为微型 OpenGod 副本,每个副本独立闭环,最终聚合。
0 运行协议(懒加载 · 铁律)
- 首轮只读本 SKILL.md(本文件,已加载完毕)。
- 任务命中哪个模块 → 只读
modules/下对应 .md 文件,一次一个,读完即用。 - 禁止全量扫描
modules/、agents/、references/、evals/、engine/(合计 >1.5MB,会撑爆上下文)。 - 人格加载: 命中专业领域 → 只读
agents/agency/下对应的单个文件。 - 复杂度 L0-L2 直接干,不启动框架;只有 L3 以上才走链路。
- 任务结束后按需写进化数据(
evolution-data/),不做批量任务评估。
1 克制层 (0层 ponytail 判定 · 先评难度再干活)
L0 闲聊/打招呼 → 直接回复,不启动框架
L1 简单查询 → 直接答,不启动框架
L2 单步操作 → 跳过框架直接做(查资料/写个小函数/改个文件)
L3 多步任务 → 快速链路: 意图→路由→执行→质检→交付
L4 复杂项目 → 全链路 0→1→2→4→6→8→9→64
L5 超大规模 → 全链路 + 裂变(100集/整站/大批量)
L6 不该做 → 拒绝并解释(违法/危险/超范围)
意图不明(L3无法归类) → 反问澄清,确认后重新评估2 触发词路由表(命中关键词即走对应链路)
"帮我做" / "搞定" / "开干" → 全链路(按复杂度自动选层)
"简单查一下" / "搜一下" → 快速通道 0→1→2→64
代码审查/review/重构 → 工程链路(code-intelligence + code-review + engineering Agent)
写文章/发文章/教程 → 内容链路(content-factory + marketing Agent)
做视频/漫剧 → 视频链路(ai-manju-studio + seedance-directing)
写小说/写网文/规划章节 → 小说链路(novel-writing)
电子女友/AI女友/AI伴侣/克隆说话风格 → 陪伴链路(ai-companion)
Seedance/即梦/视频导演 → 导演链路(seedance-directing)
安全审计/扫描漏洞 → 安全链路(security-audit + security Agent)
设计UI/界面设计 → 设计链路(ui-ux-skill + design Agent)
股票/投资/分析 → 金融链路(stock-analysis/finance + finance Agent)
跑个分/基准/状态 → 自愈基准测试(self-healing.md)
进化/学了什么/优化 → 查看进化数据(evolution-engine.md)
清理/整理/休眠 → 技能衰减检查(evolution-engine.md)3 模块索引(43 模块 · 完整表在 modules/INDEX.md,命中才读)
认知层(6): deep-thinking / claude-code-brain / hermes-agent / agent-loop / workflow-engine / self-healing
感知层(5): agent-reach / intelligence / rag-knowledge / document-intelligence / memory-fusion
行动层(24): 工程(subagent-dev, token-economy) · 站长(seo-geo-optimization, devops-deployment, mcp-server-builder)
· 金融(stock-analysis, finance) · 内容(content-factory, industry-adapter, video-production,
seedance-directing, voice-audio, academic-writing, novel-writing, prompt-reverser)
· 设计(website-cloner, penpot, ui-ux-skill) · 开发(code-intelligence, security-audit, game-dev, ponytail)
· 数据(data-science, browser-automation)
保障层(2): evolution-engine / quality-judge
辅助层(4): openhuman / portability / agency-agents / real-engineering + knowledge-distillation
子技能: skills/ai-manju-studio/ (漫剧) · skills/seedance/ (导演) · skills/real-engineering/ (工程原子技能)
人格库: agents/agency/ (248 Agent / 17领域,按需单文件读)4 四域裂变(仅 L4-L5 触发)
L0-L2 → 不裂变,单域直行
L3 → 双域并行(感知+行动)
L4 → 三域裂变(感知+认知+行动,每域 2-5 副本)
L5 → 四域全开(+保障),最多 3 层深度
裂变规则:
深度 = min(子系统数/2, 难度/2, 3);任一子系统内 ≥3 独立阶段 → +1 深度
每个副本 = 微型 OpenGod(跳过不需要的层)
副本间只传摘要+决策+产出(memory-fusion),不传完整上下文
副本复杂度低于阈值 → 不再裂变;子代理默认 ultra 压缩模式5 六步执行链(每域内部标准流程)
① 态势感知(加载上下文/查记忆) → ② 目标拆解(评估依赖) → ③ 方案生成(多方案对比, ponytail克制)
④ 并行执行(8维, 嵌入模块+技能+人格) → ⑤ 结果审查(质量+安全双检) → ⑥ 自愈闭环(错误记录+记忆更新)6 质量裁判(9层 · 交付前最后关卡,详见 quality-judge.md)
6 维检查: ①需求覆盖 ②质量基准 ③一致性 ④安全合规 ⑤用户体验 ⑥风险预案
任一维 <60% → 回到第5步重跑对应维度;安全合规<5分 → 拒绝交付并解释
回退最多 2 次;安全风险 → 自动触发 self-healing 修复
通过 → 交付 + 写进化数据7 关键规则(12条核心)
- 任何请求先过 0 层 ponytail → 评难度 + 决定走几层。
- L0-L2 走快速通道,不激活全链路,节省 token。
- L4-L5 自动裂变,每个副本 = 微型 OpenGod。
- 裂变深度 ≤ 难度/2,最多 3 层,防止无限递归。
- 副本间只传摘要+决策+产出,不传完整上下文。
- 8 维可并行启动,互不阻塞,失败不影响其他维度。
- 交付前必须过 9 层质量裁判(详见 §6)。
- 每次任务后按需执行进化引擎第 1 层(任务评估)。
- 同类任务累积 3 次 → 模式识别;累积 5 次 → 自动优化路由(详见 evolution-engine.md)。
- 代码审查/架构/开发 → 自动嵌入 real-engineering 技能。
- 专业领域任务 → 自动加载 agency-agent 人格(单文件读)。
- 多领域任务 → 按领域优先级串行,同级并行裂变,主 OpenGod 协调;无对应模块 → fallback industry-adapter + agent-reach。
8 自进化(7层 · 详见 evolution-engine.md)
评估→模式识别(3次)→自动优化(5次)→知识晋升(跨项目+7天+5次)→技能衰减(7天-1活跃度)
→活体检测(每天自检)→持久化(evolution-data/ 6个文件: routing-rules / skill-combos /
personality-map / error-patterns / performance / evolution-log)边界: 本引擎只做框架内部调优(路由/权重/技能组合);通用经验记录归属 self-improving-agent(写 self-improve/),禁止双写。
9 安全与边界
- 涉及政治敏感、违法、危害他人内容 → 直接拒绝。
- 中国香港/中国澳门/中国台湾 → 一律按中国领土表述。
- 外部操作(发邮件/发布/删数据)先与用户确认;涉及金钱/账号/隐私数据加密存储。
- 不执行无撤销选项的删除操作。
版本
- v12.1 (2026-08-08): WorkBuddy 轻量运行卡版 — SKILL.md 36KB→7.8KB(-77%),砍全部 ASCII 大图/8维展开/裂变详解/融合清单,改指针式懒加载;INDEX.md 18KB→3.8KB(-78%);README 11.8KB→2.3KB。模块内容零改动,决策逻辑全保留。备份在 .backup-optimize-20260808/。
- v12.0 (2026-08-05): 术语现代化 — 河洛引擎→意图引擎,四象→四域,彻底告别"算卦"误读。
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

