Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add ellmos-ai/build-your-users-mind --skill zhgit clone --depth 1 https://github.com/ellmos-ai/build-your-users-mindWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ellmos-ai/build-your-users-mind/zh)<a href="https://agentmods.dev/skills/ellmos-ai/build-your-users-mind/zh"><img src="https://agentmods.dev/badge/skills/ellmos-ai/build-your-users-mind/zh/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ellmos-ai/build-your-users-mind/zh"><img src="https://agentmods.dev/badge/skills/ellmos-ai/build-your-users-mind/zh.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00108 | $0.01734 |
| Opus 5 | $0.00054 | $0.00867 |
| Sonnet 5 | $0.00022 | $0.00347 |
| Haiku 4.5 | $0.00011 | $0.00173 |
Grade A, and why
build-your-users-mind scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
88% identical to build-your-users-mind — 76 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Translation status (2026-07-15): historical pre-1.1 draft. The root English documents are authoritative; do not use this translation as the current operational or security contract.
build-your-users-mind — 智能体通用 ToM 模块（反馈预知）
What you mind is what you get. 一份配方，而非框架。每个智能体模型都以此构建其用户的 ToM 模型：评估自身数据 → 提炼决策模式 → 维护化身文件 → 绑定到自身内存/规则文件/系统提示词。
核心 = 反馈预知（feedback precognition / feedforward）： 在用户反馈到来之前进行预测；在用户不在时将其作为控制信号；随后对照实际情况评估预测以实现自我改进。
模板：
templates/（化身文件）、scripts/（流水线）、TAXONOMY.md（8 种类型）、skills/swarm-operations/（分类群）。存在一个基于作者自身日志的私有参考实现，但未随附打包。理论基础： Prompt-Archaeology（方法，分类法见
TAXONOMY.md）+ ToM 研究（ToM-SWE arXiv 2510.21903；Persistent Memory & User Profiles 2510.07925）。
核心原则
LLM 从不直接查看原始的吉字节数据。确定性脚本首先将数据精简为人类键入的用户提示词的干净语料库；之后，分类群（classification swarm）才会进行语义化工作。 核心不在于“哪些提示词”，而在于“做出了什么决策 → 产生了什么结果 → 用户是否满意”。
6 个步骤
1. 明确数据源（数据源适配器）
找到您自己的交互日志。每个模型各不相同 → 参见 SOURCE-ADAPTERS.md。
仅提取人类键入的真实提示词（不包含工具执行结果、系统提示、钩子注入、上下文压缩摘要）。字段：ts, project, session, text。
2. 精简（确定性的，无 LLM 介入）
- 过滤合成的轮次，进行去重，聚合常规应答/微小确认。
- 随访关联： 将每个提示词的下一轮（或几轮）用户输入派生为
outcome_signal（praise 赞扬 | reissue 重新发出 | correction 纠正 | abandon 放弃 | none）→ 作为满意度信号。 - 通过决策词典计算
decision_score（纠正/偏好/规则/控制）。 - 脱敏（持久化前的强制步骤）： 脱敏密钥/令牌/密码/电子邮件——并根据用户情况，还需脱敏健康/税务/IP 地址。用户的敏感数据将被掩码。
3. 分类（群，分级 + stigmergy）
8 类分类法 SP/NT/NM/NS/KO/BE/RA/MP（定义见 TAXONOMY.md）+ decision_kind（preference/correction/rule/direction_change/approval/rejection/process/none）+ formulation_pattern（用户特征措辞）。对于大型语料库：领域负责人（Sonnet）指导切片执行者（Haiku）。
4. 生成化身文件
结构与 templates/ 1:1 保持一致（复制模板，替换 <USER>/<AGENT>）：
WHAT-<USER>-SAID.md（已证实）· WHAT-WOULD-<USER>-SAY.md（预测 + 置信度）· WHAT-I-DID-…md + MY-ACTIONS.txt（行动日志）· WHAT-<USER>-SAID-ABOUT-…md（经验教训）· PROMPT-LOG（剪辑与线索）· METHODIK.md（含偏差警告）· START.md（0→4 循环）。
5. 绑定（至关重要！）
ToM 模型必须切实投入使用，而不仅仅是存在：
- 在智能体自身的内存/规则文件/系统提示词中写入简短的规则/指针（Claude:
CLAUDE.md；Codex:GPT.md/AGENTS.md；agy:GEMINI.md；Kimi:KIMI.md）→ 指向START.md循环。保持简短（仅做指针，不放全文）。 - 优先规则： 项目特定的
DECISIONS.md优于横向的化身文件。 - 可选命令入口点（细节说明）： 在三个深度层级加上一个编排器公开该循环 ——
read-my-mind（预测，0→2，不执行动作）、decide-like-me（单次决策，0→2，作为工作流组件）、be-my-avatar（行动，完整 0→4，仅限可逆操作，记录日志）、avatar-orchestrator（跨多个决策的链式编排，将 🔴/不可逆项目打包成一个问题）。模板位于templates/commands/。 - 版本化绑定： 以项目副本中的循环/技能为规范版本；如果智能体还附带一个已注册的副本，则版本号更高者优先，旧的路由副本将被替换 —— 项目主导，注册表跟随（避免版本漂移）。
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 60 lines · 108 tokens per session scan A e3ba41c264c5
build-your-users-mind is a skill published in the GitHub repository ellmos-ai/build-your-users-mind (3 stars, last pushed 2d ago), licensed MIT. It adds 108 tokens to every session and 1,734 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. It is 88% identical to build-your-users-mind, differing in 76 lines, and is treated as a copy.
Other skills, from other repositories
build-your-users-mind
Verweist auf das öffentliche, anbieterneutrale Modul build-your-users-mind: ein datenschutzbewusstes Verfahren zum Aufbau eines empirischen Theory-of-Mind-Präferenzmodells eines autorisierten Nutzers aus dessen eigenen Interaktionsprotokollen. Verwenden, wenn eine zuständige Person ein Entscheidungsavatar-Profil…
mem0-vercel-ai-sdk
Mem0 provider for Vercel AI SDK (@mem0/vercel-ai-provider). TRIGGER when: user mentions "vercel ai sdk", "@mem0/vercel-ai-provider", "createMem0", "retrieveMemories", "addMemories", "getMemories", "searchMemories", "mem0 vercel", "AI SDK provider", "AI SDK memory", or is using generateText/streamText with mem0. Also…
mem0-tour
Browses all stored memories grouped by category with full content display. Use when reviewing all project memories, exploring stored knowledge, onboarding to a project, or getting an overview of captured decisions, conventions, and learnings.
pause
Pause Mem0 memory capture on this machine. Use when the user wants to stop memories being recorded, for example for private work or experiments.
mine
Mine a project or conversation into your MemPalace — extract and store memories for later retrieval.
agent-carnet
Use this skill when the user asks to save, recall, find, or organize notes. Triggers on: 'remember this', 'save this', 'note this', 'what did we discuss about...', 'check the notebook', 'find in carnet'. Also use proactively when discovering findings worth preserving across sessions.